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
231707056	pes2o/s2orc	v3-fos-license	Review on a Privacy-Preserving and Efficient k-Nearest Neighbor Query and Classification Scheme Based on k-Dimension Tree for Outsource Data – Cloud computing technology has attracted the attention of researchers and organizations due to its computing power, efficiency and flexibility. Using cloud computing technology to analyze outsourced data is become a new data utilization model. However, due to the severe security risks that appear in cloud computing, most organizations now encrypt data before outsourcing data. Therefore, in recent years, many works on the k-Nearest Neighbor (denoted by k-NN) algorithm for encrypted data has appeared. However, two main problems in existing current research are either the program is not secure enough or inefficient. In this paper, based on the existing problems, we have designed a non-interactive privacy-preserving k-query and classification scheme. Our proposed scheme uses two existing encryption schemes: Order Preserving Encryption and the Parlier cryptosystem, to preserve the privacy of encrypted outsourced data, data access patterns, and the query record, and utilizes the encrypted the k-dimensional tree (denoted by kd-tree) to optimize the traditional k-NN algorithm. Our proposed scheme aim to achieve high query efficiency while ensuring data security. Extensive experimental results prove that this scheme is almost close to the scheme using plaintext data and the existing non-interactive encrypted data query scheme in terms of classification accuracy. The query runtime of our scheme is higher than the existing non interactive k-NN query scheme. INTRODUCTION Nowadays, machine learning and cloud computing have been widely used. Machine learning can mine hidden knowledge or patterns from massive data and is one of the most attractive technologies. The K-Nearest Neighbor (denoted by K-NN) algorithm is one of classic machine learning algorithms, which can find the nearest k points from a large data set based on the test object. It has been used in many studies, such as pattern recognition, Location-Based Services, DNA sequencing, online recommendation systems, and data analysis, etc. KNN firstly computes similarities between input query and each data in dataset (Compute Similarity), converts the similarities in bitwise shared representation (Bit-Decomposition), and selects K data with the highest similarities (PE-FTK). Among the sub protocols. Update the feature extractor of a KNN model. Train a student model in the public Domain. 1. Update the feature extractor for private-KNN: We initialize the feature extractor with a public extractor -Histogram of Oriented Gradient (HOG) features. We use the neural network of the last iteration student model (except for the last softmax layer) to update the feature extractor, in the next iteration. Note that this interactive scheme will iteratively refine the feature embedding used by KNN without using any exclusive information. 2. Train a student model: When the feature extractor of Private-KNN is updated, we train a student model by labeling a limited number of student queries (the public data) with pseudolabels. For each student query, we first generate a random subset from the entire private domain, and then pick the k nearest neighbors among the subset. The pseudo-label is generated with private voting of k neighbors, and the detailed aggregation process can be found in the main paper. II-OBJECTIVES The Private-KNN is the data-efficient algorithm for differentially private (DP) deep learning under the knowledge transfer framework. It represents the practical solution that addresses this important problem scales to larger models while preserving theoretically meaning full DP guarantee. III-RELATED WORK Y. Du, " Privacy-aware run query processing on location-based services in Mobile Data Management , Reverse K nearest neighbors query processing: experiments and analysis," VLDB Endowment, vol. Zhu et al. [2] improved the scheme of Wong et al. to provide privacy-preserving k-NN query. However, the data owner has to participate in the query process, however the scheme lacks a rigorous security proof. Hu et al. [3]proposed a k-NN query scheme based on privacy homomorphism encryption scheme. In their scheme, the k-NN query on the encrypted data is achieved by holomorphic properties. How one ensured an individual"s privacy regarding his location and spatiotemporal behavioral patterns was proposed by Ho et al. (2016) through differential privacy mechanism which assumes that data trajectory is secure and users can only query knowledge derived from it. IV-TECHNIQUES In the paper, use UDA to train the student model for SVHN and CIFAR-10 tasks, which allows us to save the privacy budget with a limited number of student queries. RKNN Query Answer Retrieval Algorithm used it. The Encryption technique is Privacy Preserving: In this scheme, the cloud can't obtain the value of any data (including query data records) because we International Journal of Innovations in Engineering and Science, www.ijies.net use two encryption schemes, the OPE and Paillier cryptosystem, to encrypt the raw data and execute the k-NN algorithm. As mentioned earlier, the honest-but-curious cloud server cannot know any information about the encryption class, or even how many different classes in the scheme. And all intermediate results and the result of the query are encrypted data, and the cloud server cannot learn their plaintext data. Therefore, the encryption technique is privacy-preserving. 2. The Encrypted Data Comparison Protocol is Privacy-Preserving: The data comparison protocol is a key component of the k-query algorithm. As men-tined earlier, the comparison protocol is used in both the ki-tree technique and the comparison of distances. In our proposed scheme, we implemented the k-NN algorithm with the stateless OPE scheme presented in 2011. From this protocol, we can only obtain a comparison of two encrypted data E(a) and E(b) with-out leaking any plaintext values of a and b. Therefore, the cloud server cannot directly get the plaintext data content. Similarly, Nor can unauthorized users. Thus, this comparison protocol is privacy-preserving. V-DATA DESCRIPTION In this paper, three schemes are used to implement the k-NN algorithm: the classical plaintext k-NN classier, the scheme in and our proposed scheme. In the scheme we proposed and the scheme proposed in, a stateless OPE scheme and the Parlier holomorphic encryption scheme are used to encrypt data, including outsourced data and query records of authorized users. To be specific, the two schemes adopt the OPE scheme to encrypt the property values of the data and use the Parlier encryption scheme to encrypt the class labels of the data. Moreover, in the process of data query and classification, the comparison protocol and the computation protocol are executed between encrypted data, so the privacypreservation of all data is achieved. Therefore our proposed scheme and the scheme proposed in have the same security. VI-CONCLUSSION In this paper , proposed two novel solutions RKNN-HG and RKNN-HRT to answer private RKNN queries without disclosing any information about the location of query point. Our solutions utilize Private Information Retrieval (PIR) mechanism to request data from an untrusted database server without the server learning about retrieved data or the query source. We evaluated our methods extensively using real datasets studying the effect of several parameters on computational cost and data overhead size. Our results show the efficiency and effectiveness of our solutions. Our future work includes extending our solutions for Bichromatic RKNN queries and moving object queries for spatial crowdsourcing applications while protecting the location privacy of the participants.	2021-01-26T11:23:37.913Z	2021-01-11T00:00:00.000	{ "year": 2021, "sha1": "5199e9c9174aa655f3a60dbf852564b5aaf664dc", "oa_license": null, "oa_url": "https://doi.org/10.46335/ijies.2020.5.12.4", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "5199e9c9174aa655f3a60dbf852564b5aaf664dc", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Computer Science" ] }
49586397	pes2o/s2orc	v3-fos-license	Affine Extensions of Integer Vector Addition Systems with States We study the reachability problem for affine $\mathbb{Z}$-VASS, which are integer vector addition systems with states in which transitions perform affine transformations on the counters. This problem is easily seen to be undecidable in general, and we therefore restrict ourselves to affine $\mathbb{Z}$-VASS with the finite-monoid property (afmp-$\mathbb{Z}$-VASS). The latter have the property that the monoid generated by the matrices appearing in their affine transformations is finite. The class of afmp-$\mathbb{Z}$-VASS encompasses classical operations of counter machines such as resets, permutations, transfers and copies. We show that reachability in an afmp-$\mathbb{Z}$-VASS reduces to reachability in a $\mathbb{Z}$-VASS whose control-states grow linearly in the size of the matrix monoid. Our construction shows that reachability relations of afmp-$\mathbb{Z}$-VASS are semilinear, and in particular enables us to show that reachability in $\mathbb{Z}$-VASS with transfers and $\mathbb{Z}$-VASS with copies is PSPACE-complete. We then focus on the reachability problem for affine $\mathbb{Z}$-VASS with monogenic monoids: (possibly infinite) matrix monoids generated by a single matrix. We show that, in a particular case, the reachability problem is decidable for this class, disproving a conjecture about affine $\mathbb{Z}$-VASS with infinite matrix monoids we raised in a preliminary version of this paper. We complement this result by presenting an affine $\mathbb{Z}$-VASS with monogenic matrix monoid and undecidable reachability relation. Introduction Vector addition systems with states (VASS) are a fundamental model of computation comprising a finite-state controller with a finite number of counters ranging over the natural numbers. When a transition is taken, a counter can be incremented or decremented provided that the resulting counter value is greater than or equal to zero. Since the counters of a VASS are unbounded, a VASS gives rise to an infinite transition system. One of the biggest advantages of VASS is that most of the standard decision problems such as configuration reachability and coverability are decidable [KM69,May84,Kos82,Ler12]. Those properties make VASS and their extensions a prime choice for reasoning about and modelling concurrent, distributed and parametrised systems, see e.g. the recent surveys by Abdulla and Delzanno [AD16,Del16]. In order to increase their modelling power, numerous extensions of plain VASS have been proposed and studied in the literature over the last 25 years. Due to the infinite-state nature of VASS, even minor extensions often cross the undecidability frontier. For example, while in the extension of VASS with hierarchical zero-tests on counters both reachability and coverability remain decidable [Rei08,Bon13], all important decision problems for VASS with two counters which can arbitrarily be tested for zero are undecidable [Min67]. Another example is the extension of VASS with reset and transfer operations. In a reset VASS, transitions may set a counter to zero, whereas transfer VASS generalize reset VASS and allow transitions to move the contents of a counter onto another. While it was initially widely believed that any extension of VASS either renders both reachability and coverability undecidable, reset and transfer VASS have provided an example of an extension which leads to an undecidable reachability [AK76] yet decidable coverability problem [DFPS98]. Nevertheless, the computational costs for those extensions are high: while coverability is EXPSPACE-complete for VASS [Lip76,Rac78], it becomes Ackermann-complete in the presence of resets and transfers [Sch10,FFSPS11]. For practical purposes, the extension of VASS with transfers is particularly useful since transfer VASS allow for reasoning about broadcast protocols and multithreaded non-recursive C programs [EN98,KKW14]. It was already observed in [EN98] that transfer VASS can be viewed as an instance of so-called affine VASS. An affine VASS is a generalization of VASS with transitions labelled by pairs (A, b), where A is a d × d matrix over the integers and b ∈ Z d is an integer vector. A transition switches the control-state while updating the configuration of the counters v ∈ N d to A · v + b, provided that A · v + b ≥ 0; otherwise, the transition is blocked. Transfer VASS can be viewed as affine VASS in which the columns of all matrices are d-dimensional unit vectors [EN98]. Due to the symbolic state-explosion problem and Ackermann-hardness of coverability, standard decision procedures for transfer VASS such as the backward algorithm [ACJT96] do not per se scale to real-world instances. In recent years, numerous authors have proposed the use of over-approximations in order to attenuate the symbolic state-explosion problem for VASS and some of their extensions (see, e.g., [ELM + 14, ALW16,BH17]). Most commonly, the basic idea is to relax the integrality or non-negativity condition on the counters and to allow them to take values from the non-negative rational numbers or the integers. The latter class is usually referred to as Z-VASS, see e.g. [HH14]. It is easily seen that if a configuration is not reachable under the relaxed semantics, then the configuration is also not reachable under the standard semantics. Hence, those state-space over-approximations can, for instance, be used to prune search spaces and empirically drastically speedup classical algorithms for VASS such as the backward-algorithm. In this paper, we investigate reachability in integer over-approximations of affine VASS, i.e., affine VASS in which a configuration of the counters is a point in Z d , and in which all transitions are non-blocking. Subsequently, we refer to such VASS as affine Z-VASS. Main contributions. We focus on affine Z-VASS with the finite-monoid property (afmp-Z-VASS), i.e. where the matrix monoid generated by all matrices occurring along transitions in the affine Z-VASS is finite. By a reduction to reachability in Z-VASS, we obtain decidability of reachability for the whole class of afmp-Z-VASS and semilinearity of their reachability relations. In more detail, we show that reachability in an afmp-Z-VASS can be reduced to reachability in a Z-VASS whose size is polynomial in the size of the original afmp-Z-VASS and in the norm of the finite monoid M generated by the matrices occurring along transitions, denoted by M . For a vast number of classes of affine transformations considered in the literature, M is bounded exponentially in the dimension of the matrices. This enables us to deduce a general PSPACE upper bound for extensions of Z-VASS such as transfer Z-VASS and copy Z-VASS. By a slightly more elaborated analysis of this construction, we are also able to provide a short proof of the already known NP upper bound for reset Z-VASS [HH14]. We also show that a PSPACE lower bound of the reachability problem already holds for the extension of Z-VASS that only use permutation matrices in their transition updates. This in turn gives PSPACE-completeness of interesting classes such as transfer Z-VASS and copy Z-VASS. Finally, we show that an affine Z-VASS that has both transfers and copies may not have the finite-monoid property, and that the reachability problem for this class becomes undecidable. We complement this result by investigating the case of monogenic classes, i.e. classes of monoids with a single generator. We show that although reachability can still be undecidable for an affine Z-VASS with a monogenic matrix monoid, there exists a monogenic class without the finite-monoid property for which reachability is decidable. All complexity results obtained in this paper are summarized in Figure 1, except for the undecidability of general monogenic classes as it is a family of classes rather than one class. Related work. Our work is primarily related to the work of Finkel and Leroux [FL02], Iosif and Sangnier [IS16], Haase and Halfon [HH14], and Cadilhac, Finkel and McKenzie [CFM12,CFM13]. In [FL02], Finkel and Leroux consider a model more general than affine Z-VASS in which transitions are additionally equipped with guards which are Presburger formulas defining admissible sets of vectors in which a transition does not block. Given a sequence of transitions σ, Finkel and Leroux show that the reachability set obtained from repeatedly iterating σ, i.e., the acceleration of σ, is definable in Presburger arithmetic. Note that the model of Finkel and Leroux does not allow for control-states and the usual tricks of encoding each control-state by a counter or all control-states into three counters [HP79] do not work over Z since transitions are non-blocking. Iosif and Sangnier [IS16] investigated the complexity of model checking problems for a variant of the model of Finkel and Leroux with guards defined by convex polyhedra and with control-states over a flat structure. Haase and Halfon [HH14] studied the complexity of the reachability, coverability and inclusion problems for Z-VASS and reset Z-VASS, two submodels of the affine Z-VASS that we study in this paper. In [CFM12,CFM13], Cadilhac, Finkel and McKenzie consider an extension of Parikh automata to affine Parikh automata with the finite-monoid restriction like in our paper. These are automata recognizing boolean languages, but the finite-monoid restriction was exploited in a similar way to obtain some decidability results in that context. We finally remark that our models capture variants of cost register automata that have only one + operation [AR13,AFR14]. Structure of the paper. We introduce general notations and affine Z-VASS in Section 2. In Section 3, we give the reduction from afmp-Z-VASS to Z-VASS. Subsequently, in Section 4 we show that afmp-Z-VASS have semilinear reachability relations and discuss semilinearity of affine Z-VASS in general. In Section 5, we show PSPACE and NP upper bounds of the reachability problem for some classes of afmp-Z-VASS; and in Section 6 we show PSPACEhardness and undecidability results for some classes of affine Z-VASS. In Section 7, we show that reachability is undecidable for monogenic affine Z-VASS and remains decidable for a specific class of infinite monoids. Some concluding remarks will be made in Section 8. Preliminaries General notation. For n ∈ N, we write [n] for the set {1, 2, . . . , n}. For every x = (x 1 , x 2 , . . . , x d ) ∈ Z d and every i ∈ [d], we define x(i) def = x i . We denote the identity matrix and the zero-vector by I and 0 in every dimension, as there will be no ambiguity. For x ∈ Z d and A ∈ Z d×d , we define the max-norm of x and A as where A i denotes the i th column of A. We naturally extend this notation to finite sets, i.e. G def = max{ A : A ∈ G} for every G ⊆ fin Z d×d . We assume that numbers are represented in binary, hence the entries of vectors and matrices can be exponential in the size of their encodings. Affine Integer VASS. An affine integer vector addition system with states (affine Z-VASS) is a tuple V = (d, Q, T ) where d ∈ N, Q is a finite set and T ⊆ Q × Z d×d × Z d × Q is finite. Let us fix such a V. We call d the dimension of V and the elements of Q and T respectively control-states and transitions. For every transition t = (p, A, b, q), we define src(t) def = p, tgt(t) def = q, M (t) def = A and ∆(t) def = b, and let f t : Z d → Z d be the affine transformation defined by f t (x) = A · x + b. The size of V, denoted \|V\|, is the number of bits used to represent d, Q and T with coefficients written in binary. For our purposes, we formally define it in a crude way as \|V\| def = d + \|Q\| + (d 2 + d) · \|T \| · max(1, log( T + 1) ) where We naturally extend − → to sequences of transitions as follows. For every w = w 1 · · · w k ∈ T k and p(u), q(v) ∈ Q × Z d , we write p(u) w − → q(v) if either k = 0 (denoted w = ε) and p(u) = q(v), or k > 0 and there exist p 0 (u 0 ), p 1 (u 1 ), . . . , p k (u k ) ∈ Q × Z d such that , then we say that w is a run from p(u) to q(v), or simply a run if the source and target configurations are irrelevant. We also say that w is a path from p to q, and if p = q then we say that w is a cycle. Figure 1. Classification of the complexity of reachability in affine Z-VASS in terms of classes of matrices. The rectangular regions below and above the horizontal dashed line correspond to classes of matrices with finite and infinite monoids respectively. The rectangular green dotted region and the elliptical red striped region correspond to the classes where reachability is decidable and undecidable, respectively. The elliptical blue region and the orange elliptical region correspond to the classes where reachability is NP-complete and PSPACE-complete respectively. The term "C J -Z-VASS" refers to the specific monogenic class of infinite monoids that will be defined in Section 7.1. M, is the monoid generated by M (V), i.e. it is the smallest set that contains M (V), is closed under matrix multiplication, and contains the identity matrix. We say that a matrix A ∈ N d×d is respectively a (i) reset, (ii) permutation, (iii) transfer, (iv) copyless, or (v) copy matrix if A ∈ {0, 1} d×d and (i) A does not contain any 1 outside of its diagonal; (ii) A has exactly one 1 in each row and each column; (iii) A has exactly one 1 in each column; (iv) A has at most one 1 in each column; (v) A has exactly one 1 in each row. Analogously, we say that V is respectively a reset, permutation, transfer, copyless, or copy Z-VASS if all matrices of M (V) are reset, permutation, transfer, copyless, or copy matrices. The monoids of such affine Z-VASS are finite and respectively of size at most 2 d , d!, d d , (d + 1) d and d d . Copyless Z-VASS correspond to a model of copyless cost-register automata studied in [AFR14] (see the remark below). If M (V) only contains the identity matrix, then V is simply called a Z-VASS. A class of matrices C is a union d≥1 C d where C d is a finitely generated, but possibly infinite, submonoid of N d×d for every d ≥ 1. We say that V belongs to a class C of Z-VASS if M V ⊆ C. If each C d is finite, then we say that this class of affine Z-VASS has the finite-monoid property (afmp-Z-VASS). For two classes C and C we write C + C to denote the smallest set D = d≥1 D d such that D d is a monoid that contains both C d and C d for every d ≥ 1. Note that this operation does not preserve finiteness. For example if C and C are the classes of transfer and copy matrices, respectively, then C + C is infinite (see Figure 2 and Section 6). We say that a class C = d≥1 C d is nonnegative if C d ⊆ N d×d for every d ≥ 1. We say that an affine Z-VASS V is nonnegative if M V belongs to some nonnegative class of matrices. Note that the classes of reset, permutation, transfer, copyless and copy matrices are all nonnegative, respectively. We discuss the Z-VASS V in Figure 2 to give some intuition behind the names transfer and copy Z-VASS. The transition from p to q is a copy transition and the transition from q to p is a transfer transition. Notice that for every vector (x, y) ∈ Z 2 , we have p(x, y) − → q(x, x), i.e. the value of the first counter is copied to the second counter. Similarly, for the other transition we have q(x, y) − → p(x + y, 0), that is the value of the second counter is transferred to the first counter (resetting its own value to 0). Let A and B be the two matrices used in V. Note that (A · B) n is the matrix with all entries equal to 2 n−1 , and hence M V is infinite. Remark 2.1. The variants of affine Z-VASS that we consider are related to cost register automata (CRA) with only the + operation [AR13,AFR14] and without an output function. These are deterministic models with states and registers that upon reading an input, update their registers in the form x ← y + c, where x, y are registers and c is an integer. An affine Z-VASS does not read any input, but is nondeterministic. Thus, one can identify an affine Z-VASS with a CRA that reads sequences of transitions as words. In particular, the restrictions imposed on the studied CRAs correspond to copy Z-VASS [AR13] and copyless Z-VASS [AFR14]. Decision problems. We consider the reachability and the coverability problems parameterized by classes of matrices C: Reach C (reachability problem) Cover C (coverability problem) For standard VASS (where configurations cannot hold negative values), the coverability problem is much simpler than the reachability problem. However, for affine Z-VASS, these two problems coincide as observed in [HH14, Lemma 2]: the two problems are inter-reducible in logarithmic space at the cost of doubling the number of counters. Therefore we will only study the reachability problem in this paper. From affine Z-VASS with the finite-monoid property to Z-VASS The main result of this section is that every affine Z-VASS V with the finite monoid property can be simulated by a Z-VASS with twice the number of counters whose size is polynomial in M V and \|V\|. More formally, we show the following: Moreover, V , p and q are effectively computable from V. Proof. By Theorem 3.1, it suffices to construct, for a given afmp-Z-VASS V, the Z-VASS V and to test for reachability in V . It is known that reachability for Z-VASS is in NP [HH14]. To effectively compute V it suffices to provide a bound for M V . It is known that if \|M V \| is finite then it is bounded by a computable function (see [MS77]), and hence M V is also computable. For the remainder of this section, let us fix some affine Z-VASS V such that M V is finite. We proceed as follows to prove Theorem 3.1. First, we introduce some notations and intermediary lemmas characterizing reachability in affine Z-VASS. Next, we give a construction that essentially proves the special case of Theorem 3.1 where the initial configuration is of the form p(0). Finally, we prove Theorem 3.1 by extending this construction to the general case. It is worth noting that proving the general case is not necessary if one is only interested in deciding reachability. Indeed, an initial configuration p(v) can be turned into one of the form p (0) by adding a transition that adds v. The reason for proving the general case is that it establishes a stronger relation that allows us to prove semilinearity of afmp-Z-VASS reachability relations in Section 4. 3.1. A characterization of reachability. For every w ∈ T * , t ∈ T and u ∈ Z d , let Intuitively, for any sequence w ∈ T * , w(u) is the effect of w on u, regardless of whether w is an actual path of the underlying graph. A simple induction yields the following characterization: Testing for reachability with Lemma 3.3 requires evaluating w(u). This value can be evaluated conveniently as follows: Lemma 3.4. For every w = w 1 w 2 · · · w k ∈ T k and u ∈ Z d , the following holds: (3.1) Proof of Lemma 3.4. We prove (3.1) by induction on k. The base case follows from ε(u) = u = I · u + 0 = M (ε) · u + 0. Assume that k > 0 and that the claim holds for sequences of length k − 1. For simplicity we denote σ def = w 1 · · · w k−1 . We have: where (3.2), (3.3) and (3.4) follow respectively by definition of σw k (u), by induction hypothesis and by definition of M (σw k ). The last part of the lemma follows from applying (3.1) to w(0) and w(u), and observing that subtracting them results in w(u) − w(0) = M (w) · u. Observe that Lemma 3.4 is trivial for the particular case of Z-VASS. Indeed, we obtain w(u) = u + k i=1 ∆(w i ), which is the sum of transition vectors as expected for a Z-VASS. 3.2. Reachability from the origin. We make use of Lemmas 3.3 and 3.4 to construct a Z-VASS V = (d, Q , T ) for the special case of Theorem 3.1 where the initial configuration is of the form p(0). The states and transitions of V are defined as: The idea behind V is to simulate a path w of V backwards and to evaluate w(0) as the sum identified in Lemma 3.4. More formally, V and V are related as follows: Note that the validity of the claim completes the proof since A 0 = I and A k = M (w). It follows immediately from the definition of T that w i ∈ T for every i ∈ [k] and hence that w is a path from (q, A 0 ) to (p, A k ). By Lemma 3.3, it remains to show that w (0) = v: = w(0) (by Lemma 3.4 applied to w(0)). (b) Similarly, by Lemma 3.3, there exists a path w of V such that w (0) = v, and it suffices to exhibit a path w ∈ T * from p to q in V such that w(0) = v and M (w) = A. It is readily seen that w is a path from p to q. To prove w(0) = v and M (w) = A, Lemma 3.4 can be applied as in the previous implication. 3.3. Reachability from an arbitrary configuration. We now construct the Z-VASS V = (2d, Q , T ) of Theorem 3.1 which is obtained mostly from V . The states of V are defined as: To simplify notation, given two vectors u, v ∈ Z d we write (u, v) for the vector of Z 2d equal to u on the first d components and equal to v on the last d components. The set T consists of four disjoint subsets of transitions T simul ∪ T end ∪ T mult ∪ T final working in four sequential stages. Intuitively, these transitions allow (1) V to simulate a path w of V backwards in order to compute w(0); (2) guess the end of this path; (3) compute M (w)·u by using the fact that M (w) is stored in its control-state; and (4) guess the end of this matrix multiplication. The first set of transitions is defined as: Its purpose is to simulate T on the last d counters. The second set is defined as: and its purpose is to nondeterministically guess the end of a run in V by simply marking q. The third set is defined as: where e i is the i-th unit vector such that e i (i) = 1 and e i (j) = 0 for all i = j. The purpose of T mult is to compute A · u from the d first counters onto the d last counters. Finally, T final is defined as: and its purpose is to guess the end of the matrix multiplication performed with T mult . We may now prove Theorem 3.1: Proof of Theorem 3.1. First, note that we obtain Moreover, we have: We conclude by proving that By definition of T simul and T end , and by Proposition 3.5, it is the case that ). The transitions of T mult allow to transform (u, w(0)) into (0, w(0)+M (w)·u). Thus, using T final , we can reach the configuration p(w(0) + M (w) · u). This concludes the proof since w(u) = w(0) + M (w) · u by Lemma 3.4. ⇐) The converse implication follows the same steps as the previous one. It suffices to observe that the first part of a run of V defines the value w(0), while the second part of the run defines M (w) · u. Semilinearity of affine Z-VASS A subset of Z d is called semilinear if it is definable by a formula of Presburger arithmetic [Pre29], i.e. by a formula of FO(Z, +, <), the decidable first-order logic over Z with addition and order. Semilinear sets capture precisely finite unions of sets of the form b + N · p 1 + N · p 2 + . . . + N · p k with each p i ∈ Z d , and are effectively closed under basic operations such as finite sums, intersection and complement. Those properties make semilinear sets an important tool in many areas of computer science and find use whenever infinite subsets of Z d need to be manipulated. The results of Section 3 enable us to show that any affine Z-VASS with the finite-monoid property has a semilinear reachability relation: Let V 1 and V 2 be the affine Z-VASS illustrated in Figure 3 from left to right respectively. Note that M V 1 and M V 2 are both infinite due to the matrix made only of 1s. Moreover, the reachability relations of V 1 and V 2 are semilinear since the former can reach any target configuration from any initial configuration, and since the latter can only generate finitely many vectors due to the zero matrix. Since V 1 has a single control-state, \|M (V 1 )\| = \|M (V 2 )\| = 2 and ∆(V 2 ) = {0}, any simple natural extension of the characterization of semilinearity in terms of the number of control-states, matrices and vectors fails. It is worth noting that an affine Z-VASS with an infinite monoid may have a non semilinear reachability relation. Indeed, Figure 2 depicts a transfer + copy Z-VASS with an infinite monoid and such that {v : p(1, 1) * − → q(v)} = {(2 n , 2 n ) : n ∈ N}, which is known to be non semilinear. Moreover, this proves that even the reachability set from p(1, 1) is not semilinear. Complexity of reachability for afmp-Z-VASS In this section, we use the results of Section 3 to show that reachability belongs to PSPACE for a large class of afmp-Z-VASS encompassing all variants discussed in Section 2. Moreover, we give a novel proof to the known NP membership of reachability for reset Z-VASS. For every finite set G d ⊆ Z d×d , let G d be the monoid generated by G d . We have: Theorem 5.1. Let C = d≥1 C d be a class of matrices such that C d is finite for every d ≥ 1. It is the case that Reach C ∈ PSPACE if there exists a polynomial poly such that , and (iii) α 0 β 1 α 1 · · · β k α k is a path from p to q of length at most 2 · \|Q \| · \|T \|. For every w ∈ (T ) * , let ∆(w) def = \|w\| i=1 ∆(w i ). By Lemma 3.4 (see the remark below the proof of Lemma 3.4), we have w(u) = u + ∆(w) for every u ∈ Z d . Thus, by Lemma 3.3, checking (i), assuming (iii), amounts to testing whether e is a solution of the following system of linear Diophantine equations: x + k i=0 ∆(α i ) + ∆(β 1 ) ∆(β 2 ) · · · ∆(β k ) · e = y. (5.1) Let m def = 2 · \|Q \| · \|T \|. By Theorem 3.1, we have m ≤ 48d · \|M\| 2 · \|Q\| 2 · \|T \|. Thus, since G d is a submonoid of C d , and by assumption on C d , we have Thus, m is exponential in \|V\|. We describe a polynomial-space non deterministic Turing machine A for testing whether p (x) * − → q (y) in V . The proof follows from NPSPACE = PSPACE. Machine A guesses k ≤ \|T \|, a path π = α 0 β 1 α 1 · · · β k α k of length at most m from p to q , and e ∈ N k , and tests whether (5.1) holds for π. Note that we are not given V , but V, so we must be careful for the machine to work in polynomial space. Instead of fully constructing V and fully guessing π, we do both on the fly, and also construct ∆(α 0 ), ∆(β 1 ), . . . , ∆(β k ), ∆(α k ) on the fly as partial sums as we guess π. Note that to ensure that each β i is a cycle, we do not need to fully store β i but only its starting control-state. Moreover, note that ∆(α i ) , ∆(β i ) ≤ m · T for every i. By Theorem 3.1 and by assumption on C d , we have Hence, each α i and β i has a binary representation of polynomial size in \|V\|. By [CH16, Prop. 4], (5.1) has a solution if and only if it has a solution e ∈ N k such that Since d = 2 · d, this means that we can guess a vector e ∈ N k whose binary representation is of polynomial size, and that we can thus evaluate (5.1) in polynomial time. Corollary 5.2. The reachability problem for nonnegative afmp-Z-VASS is in PSPACE, and hence in particular for reset, permutation, transfer, copy and copyless Z-VASS. Proof. Let C = d≥1 C d be a class of nonnegative matrices. Let d ≥ 1 and let G d be a finite set of matrices such that G d ⊆ C d . By [WS91, Theorem A.2], whose proof appears in [Web87] written by one of the same authors, we have: Note that this proof applies to reset, permutation, transfer, copy and copyless classes, respectively, as they are all nonnegative. However, there is a much simpler argument for these specific classes. Indeed, their matrices all have a max-norm of a most 1 and thus their monoids contain at most 2 d 2 matrices. Theorem 5.3 [HH14]. The reachability problem for reset Z-VASS belongs to NP. Proof. Let V = (d, Q, T ) be a reset Z-VASS. The proof does not follow immediately from Theorem 3.1 because M V can be of size up to 2 d . We will analyze the construction used in the proof of Theorem 3.1, where reachability in V is effectively reduced to reachability in a Z-VASS V = (d , Q , T ). Recall that Q = (Q × M V ) ∪ (Q × M V ) ∪ Q, and thus that the size of V depends only on the sizes of Q and M V . It follows from the proof of Theorem 3.1 and Proposition 3.5 that for every run q (u, 0) * there is a corresponding run p(u) w − → q(v) in V for some w ∈ T * of length k ≥ 0. Moreover, the i th matrix occuring within the control-states of this run are is the form A i where A i = A i−1 · B for some B ∈ M V . Since M V consists of reset matrices, it holds that A 0 , A 1 , A 2 , . . . , A k is monotonic, i.e. if A i−1 has a 0 somewhere on its diagonal, then A i also contains 0 in that position. It follows that A 0 , A 1 , . . . , A k is made of at most d + 1 distinct matrices. To prove the NP upper bound we proceed as follows. We guess at most d + 1 matrices of M V that could appear in sequence A 0 , A 1 , . . . , A k . We construct the Z-VASS V as in Theorem 3.1, but we discard each control-state of Q containing a matrix not drawn from the guessed matrices. Since the constructed Z-VASS is of polynomial size, reachability can be verified in NP [HH14]. Remark 5.4. Observe that the proof of Theorem 5.3 holds for any class of affine Z-VASS with a finite monoid such that every path of its Cayley graph contains at most polynomially many different vertices. For a reset Z-VASS of dimension d, the number of vertices on every path of the Cayley graph is bounded by d + 1. Hardness results for reachability It is known that the reachability problem for Z-VASS is already NP-hard [HH14], which means that reachability is NP-hard for all classes of affine Z-VASS. In this section, we show that PSPACE-hardness holds for some classes, matching the PSPACE upper bound derived in Section 5. Moreover, we observe that reachability is undecidable for transfer + copy Z-VASS. Theorem 6.1. The reachability problem for permutation Z-VASS is PSPACE-hard. Proof. We give a reduction from the membership problem of linear bounded automata, which is known to be PSPACE-complete (see, e.g., [HU79,Sect. 9.3 and 13]). Let A = (P, Σ, Γ, δ, q ini , q acc , q rej ) be a linear bounded automaton, where: • P is the set of states, • Σ ⊆ Γ is the input alphabet, • Γ is the tape alphabet, • δ is the transition function, and • q ini , q acc , q rej are the initial, accepting and rejecting states respectively. The transition function is a mapping δ : P × Γ → P × Γ × {Left, Right}. The intended meaning of a transition δ(p, a) = (q, b, D) is that whenever A is in state p and holds letter a at the current position of its tape, then A overwrites a with b and moves to state q and to the next tape position in direction D. Let us fix a word w ∈ Σ * of length n that we will check for membership. We construct a permutation Z-VASS V = (d, Q, T ) and configurations r(u) and r (0) such that A accepts w if and only if r(u) * − → r (0). We set d def = n · \|Γ\| + 1 and associate a counter to each position of w and each letter of the tape alphabet Γ, plus one additional counter. For readability, we denote these counters respectively as x i,a and y, where i ∈ [n] and a ∈ Γ. The idea is to maintain, for every i ∈ [n], a single non zero counter among {x i,a : a ∈ Γ} in order to represent the current letter in the i th tape cell of A. The initial vector is u ∈ {0, 1} d such that u(y) = n and u(x i,a ) = 1 if and only if w i = a for every i ∈ [n] and a ∈ Γ. The invariant that will be maintained during all runs is y = i,a x i,a . The control-states of V are defined as: The purpose of each state of the form r p,i is to store the current state p and head position i of A. States of the form r a,i will be part of a gadget testing whether A is simulated faithfully. We associate a transition to every triple (p, a, i) ∈ P × Γ × The purpose of the swap is to simulate the transition of A, upon reading a in tape cell i and state p, by moving the contents from x i,a to x i,b . Note that this transition may be faulty, i.e. it can simulate reading letter a even though tape cell i contains another letter. The purpose of the vector a is to detect such faulty behaviour: if the cell i does not contain a, then more than one counter among {x i,a : a ∈ Γ} will be a nonzero counter. Recall that y = i,a x i,a . We conclude that A accepts w if and only if there exist j ∈ [n], u ∈ N d and a 1 , a 2 , . . . , a n ∈ Γ such that r q ini ,1 (u) * − → r q acc ,j (u ) and u (y) = i∈ [d] u (x i,a i ). To test whether such index j, vector u and letters a 1 , a 2 , . . . , a n exist, we add some transitions to T as illustrated on the right of Figure 4. For every i ∈ [n] and every a ∈ Γ, we add to T the transitions (r q acc ,i , I, 0, r a,1 ). For every i ∈ [n] and a ∈ Γ, we add to T the transitions (r a,i , I, b, r a,i ) where b is the vector whose only non zero components are b(x i,a ) = b(y) = −1. Moreover, if i < n, then for every a, b ∈ Γ we also add transitions (r a,i , I, 0, r b,i+1 ). Finally, for all a ∈ Γ, we also add transitions (r a,n , I, 0, r acc ). The purpose of these transitions is to guess for each i some letter a i and simultaneously decrease x i,a i and y. We do this for each i starting from 1 to n and in the end we move to the state r acc . We conclude that A accepts w if and only if r q ini ,1 (u) * − → r acc (0) in V. Corollary 6.2. The reachability problem is PSPACE-complete for permutation Z-VASS, transfer Z-VASS and copy Z-VASS. Proof. PSPACE-hardness for permutation Z-VASS was shown in Theorem 6.1, and the upper bound for transfer Z-VASS and copy Z-VASS follows from Theorem 5.1. It remains to observe that permutation matrices are also transfer and copy matrices. Proposition 6.3. The reachability problem for transfer + copy Z-VASS is undecidable, even when restricted to three counters. Proof. Reichert [Rei15] gives a reduction from the Post correspondence problem over the alphabet {0, 1} to reachability in affine Z-VASS with two counters. The trick of the reduction Figure 5. Gadget (on the right) made of copy and transfer transitions simulating the doubling transition on the left. is to represent two binary sequences as the natural numbers the sequences encode, one in each counter. If we add an artificial 1 at the beginning of the two binary sequences, then these sequences are uniquely determined by their numerical values. We only need to be able to double the counter values, which corresponds to shifting the sequences. This can be achieved using the following matrices: The only matrices used in the construction of Reichert are I, D 1 and D 2 . The last two matrices can be simulated by a gadget made of copy and transfer matrices and by introducing a third counter. This gadget is depicted in Figure 5 for the case of matrix D 1 . The other gadget is symmetric. Note that if a run enters control-state p of the gadget with vector (x, y, 0), then it leaves control-state q in vector (2x + b 1 , y + b 2 , 0) as required. Remark 6.4. The coverability problem for nonnegative affine VASS is known to be decidable in Ackermann time [FFSPS11]. Recall that coverability and reachability are inter-reducible for affine Z-VASS. Thus, Proposition 6.3 gives an example of a decision problem, namely coverability, which is more difficult for affine Z-VASS than for affine VASS. Reachability beyond finite monoids Thus far, we have shown, on the one hand, that reachability is decidable for affine Z-VASS with the finite-monoid property, and, on the other hand, that reachability is undecidable for arbitrary affine Z-VASS. This raises the question of whether there is a decidability dichotomy between classes of finite and infinite monoids, i.e. whether reachability is undecidable for every class of infinite monoids. In this section, we show that this is not the case: we exhibit a non-trivial class of infinite monoids for which affine Z-VASS reachability is decidable. In other words, the top rectangular region of Figure 1 is not equal to the red ellipse, which answers a question we left open in [BHM18]. The class of affine Z-VASS will have a particular shape, namely, the matrix monoids have a single generator. More formally, we say that a class of matrices C = d≥1 C d is monogenic if each monoid C d is generated by a single matrix. In the second part of this section we prove that reachability is in general undecidable for monogenic classes. 7.1. Decidability for a class of affine Z-VASS with infinite monoids. Let C d be the monoid generated by the (nonnegative) matrix J d ∈ N d×d whose entries are all equal to 1. Clearly, C d is infinite for every d ≥ 2 since (J d ) n is the matrix whose entries are all equal to d n . Let C J = d≥1 C d . The rest of this section is devoted to proving the following theorem: Theorem 7.1. The reachability problem Reach C J is decidable. Let V = (d, Q, T ) be an affine Z-VASS belonging to C J . We will simply write J instead of J d as d is implicit from the dimension of V. Observe that we can assume w.l.o.g. that for every transition (p, A, b, q) ∈ T either A = I or b = 0, i.e. each transition either performs a transformation of the form x ← x + b or x ← A · x. Indeed, by adding an extra state r, we can always split such a transition into two transitions (p, A, 0, r) and (r, I, b, q). We can further assume w.l.o.g. that I and J are the only matrices occurring in V. Indeed, if T contains a transition t = (p, A, b, q) where A ∈ {I, J}, then A = J n for some n ≥ 2 and b = 0. Thus, we can simply replace t by a sequence of transitions t 1 , t 2 , . . . , t n leading from p to q and such that M (t i ) = J and ∆(t i ) = 0 for every i ∈ [n]. Let T I and T J denote the (maximal) subsets of T of transitions with matrix I and J respectively. Note that T I and T J form a partition of T . We will write S − → and S * −→ to denote respectively the restriction of − → and (1) p(u) for some w ∈ T * . If w does not contain any transition from T J , then (1) holds and we are done. Thus, suppose that w contains at least one transition from T J . Let t ∈ T J be the last such transition occurring in w. Recall that, by assumption, M (t) = J and ∆(t) = 0. Therefore, we are done since there exist r, r ∈ Q and w ∈ Z d such that In order to prove that Reach C J is decidable, it suffices to show that there exist procedures to decide the two conditions of Proposition 7.2. Testing condition (1) amounts to Z-VASS reachability, which belongs to NP [HH14]. Indeed, any run restricted to T I is a run of the Z-VASS induced by T I . Thus, in the rest of the proof, we focus on showing how to test condition (2). For this purpose, let us introduce an auxiliary model. An affine one-counter Z-net is a pair (P, U ) where • P is a finite set of states, and Furthermore, for every transition t = (p, , c, q), we write p(n) The notions of runs and reachability are defined accordingly as for affine Z-VASS. These machines are a special case of one-counter register machines with polynomial updates whose reachability problem belongs to PSPACE [FGH13], i.e. we only allow the counter to be multiplied or incremented by constants, whereas the model of [FGH13] allows to update the counter by a polynomial such as x 2 or x 3 − x + 1. For every v ∈ Z d , let Consider the transitions T in the affine Z-VASS V. For every transition t ∈ T , let t be defined as: where p = src(t) and q = tgt(t). Let W = (Q, T ) be the affine one-counter Z-net obtained from V = (d, Q, T ) by keeping the same states and taking T def = {t : t ∈ T }. We write w ∈ T * to denote the (unique) sequence of transitions in W corresponding to the sequence w ∈ T * of V. Let us observe the following correspondence between V and W: Lemma 7.3. For every p, q ∈ Q, u ∈ Z d , m ∈ Z and w ∈ T * , we have p(δ(u)) Proof. The claim follows from a simple induction on \|w\|. We may now prove Theorem 7.1. Proof of Theorem 7.1. Recall that it suffices to show how to decide condition (2) of Proposition 7.2. By definition of J, this condition is equivalent to determining whether there exist r ∈ Q and n ∈ Z such that p(u) Let S = {m ∈ Z : ∃n ∈ Z (m = d · n) ∧ r∈Q r(n, n, . . . , n) As we mentioned earlier, V can be seen as a standard Z-VASS when restricted to T I . Since the reachability relation of any Z-VASS is effectively semilinear [HH14], the set S is also effectively semilinear. Thus, it suffices to test whether p(δ(u)) in W for some m = d · n ∈ S. This can be achieved by extending W with a gadget that non deterministically subtracts some element of S after executing a transition from T J . More precisely, since S is an (effectively) semilinear set of integers, it is also (effectively) ultimately periodic. Thus, it is possible to obtain a description of S = F ∪ B + a · N where F = {f 1 , f 2 , . . . , f k } and B = {b 1 , b 2 , . . . , b } are finite subsets of Z. We extend W with the gadget depicted in Figure 6. More precisely, for every transition t ∈ T J leading to r, we add a new transition leading to a gadget that either subtracts some number from F or some number from B + a · N. Note that the gadget is not "attached" directly to r as we must ensure that r is entered by a transition of T J . Hence, testing whether amounts to testing whether p(δ(u)) * − → r (0) in the new net. Since the latter can be done in polynomial space [FGH13], we are done. 7.2. Undecidability for monogenic classes. In contrast with the previous result, we prove that decidability is undecidable in general for monogenic classes: Theorem 7.4. Reachability for monogenic affine Z-VASS is undecidable. Moreover, there exists a fixed monogenic affine Z-VASS for which deciding reachability is undecidable. We show the first part of Theorem 7.4 by giving a reduction from the problem of determining whether a given Diophantine equation has a solution over the natural numbers, which is well-known to be undecidable. The second part of Theorem 7.4 follows as a corollary. Indeed, by Matiyasevich's theorem, Diophantine sets correspond to recursively enumerable sets. In particular, there exists a polynomial P such that x ∈ N is the encoding of a halting Turing machine ⇐⇒ ∃y : P (x, y) = 0. The forthcoming construction will yield a monogenic affine Z-VASS that can test "∃y : P (x, y) = 0" by nondeterministically guessing y and testing P (x, y) = 0. Hence, reachability cannot be decided for this monogenic affine Z-VASS as the above language is undecidable. Let us show the first part of Theorem 7.4. Let x 1 , x 2 , . . . , x k be variables of a given polynomial P (x 1 , x 2 , . . . , x k ). We will construct an instance of the reachability problem, for a monogenic affine Z-VASS V, such that reachability holds if and only if P (x 1 , x 2 , . . . , x k ) = 0 has a solution over N k . The affine Z-VASS will be described using the syntax of counter programs; see [Esp98, CLL + 19], where a similar syntax was used to present the VASS model. We will make use of two instructions: zero(x)? and loop. The former checks whether counter x has value 0, and the latter repeats a block of instructions an arbitrary number of times. Figure 7 gives an example of such a program together with its translation as an affine Z-VASS. Macros. Before describing the reduction, let us introduce helpful macros. First, we define macros "transfer x onto y" and "remove x from y". The former computes y = y + x and x = 0, and the latter computes y = y − x and x = 0. Both macros work under the assumption that x is initially non negative. These macros are implemented as follows: We define another macro "t = square(s)" for squaring the contents of a counter. More precisely, it computes s = t 2 and t = 0. This macro is implemented as follows: Right: an affine Z-VASS V equivalent to P, where its first and second components correspond to counters x and y respectively. The program loops are simulated by loops within the control structure of V. Note that whenever P only adds and subtracts constants from counters, the associated matrix is the identity. Since the only way V can test whether a counter equals 0 is at the end of the program via a reachability query, instruction zero(y)? merely emphasizes that counter y will never be used again. The above program starts with t and its auxiliary counters set to 0, and ends with s and the auxiliary counters set to 0. Its correctness follows by observing that (n + 1) 2 = n 2 + (2n + 1). We introduce one last macro "t = mult(s, s )" for multiplication. More precisely, it computes t = s · s and s = s = 0. Its implementation exploits the fact that 2mn = (m + n) 2 − m 2 − n 2 : // t = s · s , z = 0 The above program starts with t and its auxiliary counters set to 0, and ends with s, s and its auxiliary counters set to 0. Note that a macro "t = mult(s, c)" for multiplying by a constant c can be achieved by a simpler program: Although these programs can be implemented rather straightforwardly by an affine Z-VASS, two remarks are in order: • Affine Z-VASS do not have any native operation for testing a counter for zero. However, a counter x can be tested once via a reachability query, provided that x is left untouched after instruction zero(x)? has been invoked. Consequently, a constant number of zero-tests can be performed on a counter, provided its initial contents has been duplicated; • Instruction "transfer s onto t" at line 2 of macro "t = square(s)" destroys the contents of s which is later needed at line 9. As for zero-tests, this is not an issue provided that some counter holds a copy of s. Thus, only a constant number of squaring, and hence of multiplications, can be performed from a given counter. The construction. Let us now describe the reachability instance. The initial vector is 0, which corresponds to having all counters set to 0 at the start of the program. The target vector is also 0, which corresponds to performing zero tests on all counters. The program starts by performing a sequence of loops that guess a valuation x for which P (x) = 0 is to be tested. More precisely, a value is nondeterministically picked for each variable x i and stored in counters x 1 i , x 2 i , . . . , x n i i . The reason for having n i copies of the value is to address the two issues mentioned earlier concerning zero-tests and reusing counters within macros. The precise number of copies, n i , will be determined later. The fragment of code achieving the initialization is as follows: loop x 1 1 = x 1 1 + 1; x 2 1 = x 2 1 + 1; · · · x n 1 1 = x n 1 1 + 1 loop x 1 2 = x 1 2 + 1; x 2 2 = x 2 2 + 1; · · · x n 2 2 = x n 2 2 + 1 . . . loop x 1 k = x 1 k + 1; x 2 k = x 2 k + 1; · · · x n k k = x n k k + 1 After the initialization, we compute the value of each monomial occurring within polynomial P (x 1 , x 2 , . . . , x k ). This can be achieved using counters x j i and the multiplication macro. Let Q(x 1 , x 2 , . . . , x k ) be a monomial of degree d. We show how to proceed by induction on d. If d = 0, then this is trivial. For larger degrees, we evaluate Q without its coefficient c, and then apply macro "t = mult(s, c)". If d = 1, then we can simply transfer the appropriate counter x j i . Otherwise, Q is a product of two monomials Q and Q of smaller degrees. By induction hypothesis, we can construct three copies of both monomials Q and Q . Then, using the multiplication macro, we obtain Q. Having evaluated all monomials, we can transfer each of their values to a common counter using the transfer and remove macros depending on whether their sign is positive or negative. Finally, we test if the resulting value equals zero, which corresponds to having a solution to P (x 1 , x 2 , . . . , x k ) = 0. Before arguing correctness of the construction, let us see why the whole program can be translated as a monogenic affine Z-VASS V, i.e. using only the identity matrix and one extra matrix A. First, note that all macros have internal counters x, y and z. Every time we use a macro, we use three fresh counters, increasing the dimension of V. Second, note that the only macro that requires a matrix different from the identity is "t = square(s)", which doubles y during an assignment to z at line 6. We will construct the matrix A as follows. Suppose we want to encode one of the squaring macros. Matrix A will have the same updates for all counters denoted as z in all macros, but the vector will use constants according to this macro. That is, all coordinates for counters not occurring in this macro will be 0. In particular, counter z from this macro will be updated like in line 6, i.e. "z = z + 2y + 1", and all other counters corresponding to some other z will be updated as "z = z + 2y". Correctness. We conclude by proving correctness of the construction. If P has a solution, then it is straightforward to extract a run from V: (a) each x j i is initialized according to the solution; and (b) each loop of the program is performed the exact number times so that each zero-test holds. It remains to observe that after performing a zero-test on some counter, V does not perform any operation on this counter or performs "z = z + 2y". But if the values of y and z are equal to zero, then z will remain equal to zero after such an update. Conversely, suppose there is a reachability witness (from 0 to 0). We claim that the initialization of counters x j i provides a solution to P . To prove this, it suffices to show that every zero-test was valid. This is clear for all counters except for the z within the squaring macro. Indeed, all other counters never change their values afterwards. However, counter z is updated by "z = z + 2y". If y is non zero, then this will be detected by the zero-test on y. Otherwise, the update "z = z + 2y" never changes the value of z as required. Conclusion We have shown that the reachability problem for afmp-Z-VASS reduces to the reachability problem for Z-VASS, i.e. every afmp-Z-VASS V can be simulated by a Z-VASS of size polynomial in \|V\|, \|M V \| and M V . In particular, this allowed us to establish that the reachability relation of any afmp-Z-VASS is semilinear. For all nonnegative classes and consequently for all of the variants we studied -reset, permutation, transfer, copy and copyless Z-VASS -\|M V \| and M V are of exponential size, thus yielding a PSPACE upper bound on their reachability problems. Moreover, we have established PSPACE-hardness for all of these specific classes, except for the reset case which is NP-complete. We do not know whether an exponential bound on M V holds for any class of afmp-Z-VASS over Z d×d . We are aware that an exponential upper bound holds when M V is generated by a single matrix [IS16]; and when M V is a group then we have an exponential bound but only on \|M V \| (see [KP02] for an exposition on the group case). Finally, we have shown that there exists a (monogenic) class without the finite-monoid property for which reachability is decidable. This result was complemented by showing that reachability is undecidable in general for monogenic classes.	2018-07-07T00:25:53.504Z	2019-09-26T00:00:00.000	{ "year": 2019, "sha1": "8b5d2c23b4282068ddf4b63c9da6f555815d8dd1", "oa_license": "CCBY", "oa_url": "https://lmcs.episciences.org/7686/pdf", "oa_status": "GOLD", "pdf_src": "Arxiv", "pdf_hash": "2cc57377d0b30c87a603f44db9299cde611185e1", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Computer Science", "Mathematics" ] }
79540271	pes2o/s2orc	v3-fos-license	The prevalence and type of pulmonary involvement in ankylosing spondylitis Nader Rezaei1, Simin Almasi2*, Kazem Zamani, Ali Khalooei Assistant Professor of Pulmonary Medicine, Department of Pulmonary Medicine, Firouzgar Hospital, Iran University of Medical Science, Tehran, Iran; Assistant Professor of Rheumatology, Department of Rheumatology, Firouzgar Hospital, Iran University of Medical Sciences, Tehran, Iran ; Internal Medicine, Department of Rheumatology, Firouzgar Hospital, Iran University of Medical Science, Tehran, Iran; Assistant Professor of Community Medicine, Department of Community Medicine, School of Medicine, Kerman University of Medical Sciences, Kerman, Iran Introduction ____________________________ Ankylosing spondylitis (AS) is a chronic and inflammatory disease that primarily affects the axial skeleton, the entheses, and, less frequently, the peripheral joints.The most prevalent symptom of AS is chronic low back pain, and clinical manifestations usually include limited spinal range of motion, predominantly symmetric bilateral sacroiliitis on radiography, and positive serology for HLA-B27 in 90-95% of cases [1][2][3]. Spondylitis can be associated with extra-articular findings, particularly uveitis, and less commonly, cardiac and pulmonary disease [4][5][6].Pulmonary disease can develop as a result of restrictive changes caused by the musculoskeletal disease and changes in the lungs themselves, including interstitial, nodular, and parenchymal abnormalities [6,7].AS occasionally affects the cricoarytenoid joint, and patients can experience hoarseness, sore throat, upper airway obstruction, or respiratory failure.Furthermore, some patients present with other pulmonary changes such as apical pulmonary fibrosis through conventional radiography.Apical fibrosis is usually asymptomatic and in association with a long duration of disease [8][9][10][11].Although high-resolution computed tomography (HRCT) provides a good diagnostic method for the detection of pulmonary abnormalities [11], the exact prevalence and type of such manifestations remain unknown. Considering the significance of detecting pulmonary involvement as soon as possible, estimating the prevalence and determining the type of such abnormalities are very important.This single center crosssectional study was undertaken to provide data on the prevalence and type of AS pulmonary manifestations. In these patients, lung involvement can have a poor prognosis, and abnormal lung parenchyma causes superinfection with a number of organisms, such as aspergillus or mycobacterium tuberculosis [12,13]. There are no treatments for fibrobullous diseases in ankylosing spondylitis patients, and there are no similar opinions about the therapeutic effects of anti-TNF drugs in pulmonary involvement [14]. Patients and settings This is a cross-sectional study, and cases were selected from the registry system of the ankylosing spondylitis patients in Firouzgar Hospital in Tehran, Iran. Diagnostic criteria for all patients included the diagnostic criteria from the 1984 modified New York criteria for AS [15].Patients that attended periodic visits were entered into the study. Informed written consent was obtained from all patients prior to enrollment.Patients were excluded if they had history of asthma or COPD. BASDAI was used to determine disease severity.If BASDAI was less than 4, the disease was considered inactive; if it was equal to or more than 4, the disease was considered active.All patients were examined, and chest X-ray (CXR) and pulmonary function test (PFT) were performed; if the physical examination, CXR, or PFT results were abnormal, HRCT was performed. All lung CT scans, CXR, and PFT were evaluated by a pulmonologist, and all results were collected in the provided forms. This study could have been improved by including a larger number of patients, but the strength of this study is in the patient selection.All patients in this study fulfilled the 1984 modified New York criteria for AS, and, as is known, the specificity of this criteria is very high for the diagnosis of this disease.Therefore, the number of patients was low. All patients in this study had a high probability of ankylosing spondylitis, but not any other spondyloarthropathy. If other diagnostic criteria had been used for patients, it is possible that not all studied patients had ankylosing spondylitis, and the amount of bias would have been minimized.It should be noted that the Ethics Committee reviewed and approved this study. Statistical analysis After data was collected, it was entered into the SPSS version 16 software with T-TEST and Nonparametric test (one sample Kolmogorov-Smirnov test).Continuous data was shown as mean ± standard deviation (SD), and categorical data as numbers with percentages (n, %), and cross tabs were analyzed according to the requirements of information.Twenty patients were smokers (33.3%), and all patients were consuming less than one pack per year.Chest X-rays were taken from all patients.Ten patients had abnormalities on CXR, including reticular pattern, atelectasis, nonspecific lucency, thickening of the pleural, and elevated hemidiaphragm (Table 2).In this study, PFT was taken from all patients, and 15 (25%) patients had abnormal PFT, 11 (18.3%)patients had a restrictive pattern, and 4 (6.7%)patients had an obstructive pattern. Results ________________________________ Lung CT scans were taken of 14 patients and the results are shown in Table 3. Discussion ______________________________ Ankylosing spondylitis (AS) is one of the families of spondyloarthropathy; there is a direct correlation between the prevalence of the human leukocyte antigen HLA-B27 and ankylosing spondylitis. Reports of pulmonary disease in these patients is not high; however, it seems that many of these patients are asymptomatic. HRCT abnormalities in these patients that have been reported include unilateral or bilateral apical fibrosis, emphysema, thickening of the pleura and pleural disease, interstitial lung disease (ILD), and mycetoma. In the current study, HRCT was performed in only 14 (23.3%)patients, for ethical issues.Among these, 8 (57.1%) patients had abnormality in HRCT scan. Although not all patients in this study had an HRCT scan, for those patients who were symptomatic in history, physical exam, or had an abnormality in CXR or PFT, HRCT was performed. Of the patients who had CT scan, only 6 (42.9%) had an active disease.There is no statistically significant relationship between disease activity and HRCT scan (p=0.56). In these patients, the only statistically significant association was found between gender (p=0.01) and CRP (p=0.01) and CT findings.Complete patient information is presented in Table 4. HRCT was not requested for all patients, and this is a challenge in the current study.The Ethics Committee did not allow HRCT to be taken in asymptomatic patients. In this study, 10 (16.6%) patients had an abnormality in CXR; the most abnormal changes included reticular (3.3%) and nodular (3.3%) patterns.Only 2 (3.3%) patients had normal CXR and abnormal HRCT.In Kiris et al. ( 16), all 28 patients with AS had a normal CXR, and 18 of these patients had an abnormal HRCT.This result differs from the results of the current study.In this study, 7 out of 10 patients who had abnormal CXR were smokers, and 13 out of 50 patients who had normal CXR were smokers.The mean disease duration was 9.4 years in patients who had abnormal CXR and 10.5 years in patients with normal CXR.Thus, patients who were smokers had more abnormal changes in CXR (p=0.05). The Pulmonary Function Test (PFT) was performed for all patients; 15 (25%) patients showed abnormal PFT, 11 (18.3%)patients showed a restrictive pattern, and 4 (6.7%)patients had an obstructive pattern.Of the 15 patients who had abnormal PFT, only 5 of them were smokers.No significant association between smoking and abnormal PFT (p-value=0.38) was found.The mean of BASFI in patients who have had a restrictive pattern was 3.1, and there is a significant association between BASFI and restrictive pattern in these patients (p=0.005).In Berdal et al., as in the current study, showed that most patients with a restrictive pattern had reduced range of motion in the spine [17].Thus, preventing structural deformity can be prevented in these patients, it would be a great help to our patients.It seems that in ankylosing spondylitis (AS), the prevalence of pulmonary involvement is more particularly a restrictive pattern.Berdal et al. (2012) showed significantly impaired pulmonary function in AS patients compared to the controls and demonstrated a clear relationship between reduced spinal mobility and restrictive PF in AS patients [17]. Conclusion _____________________________ This study showed that the incidence of lung involvement in patients with AS is similar to that of other studies (16%), and in most cases, lung involvement is of the restrictive pattern.Lung involvement in these patients may be asymptomatic; thus, it seems that performing PFT and CXR would be better for all patients with AS, even if they have not had symptoms of respiratory involvement, PFT, or CXR. Table 1 . This study was conducted on 60 (M/F=51/9) patients with AS.Mean age was 38.7±10.7,with the youngest being 22 years old and the oldest 65 years old.Disease duration was 10.29±5.8years,ESR was 24.1±20.6,CRPwas8.63±15.2,BASDAIwas3±1.8 (BASDAI˂4 inactive diseases and BASDAI≥4 active diseases), and BASFI was 2.3±1.8.Other characteristics are presented in Table1.This study was conducted in accordance with the ethical principles outlined in the World Medical Association Declaration of Helsinki.Characteristic of patients Table 3 . Frequency of HRCT findings Table 4 . Characteristics of the patients who became the CT lung	2018-09-05T15:42:23.945Z	2018-04-01T00:00:00.000	{ "year": 2018, "sha1": "45ffca4162a4cc97ff68d46b3875c7b725143923", "oa_license": "CCBY", "oa_url": "http://www.rheumres.org/article_51025_f43ef7e34b197ae4598e5b42f2cc8aae.pdf", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "45ffca4162a4cc97ff68d46b3875c7b725143923", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
10772757	pes2o/s2orc	v3-fos-license	Expected Shortfall: a natural coherent alternative to Value at Risk We discuss the coherence properties of Expected Shortfall (ES) as a financial risk measure. This statistic arises in a natural way from the estimation of the"average of the 100p % worst losses"in a sample of returns to a portfolio. Here p is some fixed confidence level. We also compare several alternative representations of ES which turn out to be more appropriate for certain purposes. A four years impasse Risk professionals have been looking for a coherent alternative to Value at Risk (VaR) for four years. Since the appearance, in 1997, of Thinking Coherently by Artzner et al [3] followed by Coherent Measures of Risk [4], it was clear to risk practitioners and researchers that the gap between market practice and theoretical progress had suddenly widened enormously. These papers in fact faced for the first time the problem of defining in a clearcut way what properties a statistic of a portfolio should have in order to be considered a sensible risk measure. The answer to this question was given through a complete characterization of such properties via an axiomatic formulation of the concept of coherent risk measure. With this result, risk management became all of a sudden a science in itself with its own rules correctly defined in a deductive framework. Surprisingly enough, however, VaR, the risk measure adopted as best practice by essentially all banks and regulators, happened to fail the exam for being admitted in this science. VaR is not a coherent risk measure because it simply doesn't fulfill one of the axioms of coherence. Things are worse than some people strive to believe. The fact that for years the class of coherent measures didn't exhibit any known specimen that shared with VaR its formidable advantages (simplicity, wide applicability, universality,. . . ) led many practitioners to think that coherence might be some sort of optional property that a risk measure can or cannot display. It seemed that coherent measures belonged to some ideal world which real-world practical risk measures can only dream of. So little attention was paid to this problem that to the best of our knowledge no risk management textbook ever mentioned the fact that VaR is not coherent. This attitude means underestimating the impact of the conclusions of [3]. Writing axioms means crystallizing in a minimal number of precise statements the intrinsic nature of a concept. It is a necessary step to take in the process of translating a complex reality into a mathematical formulation. The axioms of coherence simply embody in a synthetic and essential way those features that single out a risk measure in the class of statistics of a portfolio dynamics, just like the axiom "it must be higher when air is hotter" identifies a measure of temperature out of the class of thermodynamical properties of the atmosphere. If you want to use a barometer for measuring temperature despite the fact that pressure does not satisfy the above axiom, don't be surprised if you happen to be dressed like an Eskimo in a hot cloudy day or to be wearing a swim costume in an icy sunshine. Broken axioms always lead to paradoxical, wrong results. And VaR makes no exception. Once you know which axiom is violated by VaR it is a child's play to provide examples where the assessment of risks via VaR is definitely wrong or, in other words, where higher VaR figures come from less risky portfolios [1]. In this paper and henceforth we are going to take these axioms seriously as many other groups of researchers [4,5,6,7,9,10], practitioners and regulators [11] have begun to do. To avoid confusion, if a measure is not coherent we just choose not to call it a risk measure at all. In other words, for us, the above-mentioned axioms define the concept of risk itself via the characterization of the possible operative ways to measure it 1 . This might seem a dogmatic approach but it is not. We are of course prepared to give up this definition as soon as a new different set of axioms is proposed which is more suitable to a mathematical formulation of the concept of risk measure. What we are not prepared to do anymore, after we learned the lesson of [3,4], is discussing of risk measures without even defining what "risk measure" means. We therefore promote the coherence axioms of [3] to key defining properties of any risk measure to clearly state that in our opinion speaking of non-coherent measures of risk is as useless and dangerous as speaking of non-coherent measures of temperature. In our language, the adjective coherent is simply redundant. Definition 1 (Risk Measure) Consider a set V of real-valued random variables. A function 1 Note that this was indeed the genuine motivation of [4]. Quoting from the introduction: We provide in this paper a definition of risk . . . and provide and justify a unified framework for the analysis, construction and implementation of measures of risk. VaR is not a risk measure because it does not fulfill the axiom of sub-additivity. This property expresses the fact that a portfolio made of sub-portfolios will risk an amount which is at most the sum of the separate amounts risked by its sub-portfolios. This is maybe the most characterizing feature of a risk measure, something which belongs to everybody's concept of risk. The global risk of a portfolio will be the sum of the risks of its parts only in the case when the latter can be triggered by concurrent events, namely if the sources of these risks may conspire to act altogether. In all other cases, the global risk of the portfolio will be strictly less than the sum of its partial risks thanks to risk diversification. This axiom captures the essence of how a risk measure should behave under the composition/addition of portfolios. It is the key test for checking whether a measurement of a portfolio's risk is consistent with those of its parts. For a sub-additive measure, portfolio diversification always leads to risk reduction, while for measures which violate this axiom, diversification may produce an increase in their value even when partial risks are triggered by mutually exclusive events [1]. Sub-additivity is necessary for capital adequacy requirements in banking supervision. Think of a bank made of several branches: if the capital requirement of each branch is dimensioned on its own risk, the regulator should be confident that also the overall bank capital should be an adequate one. This may however not be the case if the adopted measure violates sub-additivity since the risk of the whole bank could turn out to be much bigger then the sum of the branches' risks. Sub-additivity is an essential property also in portfolio-optimization problems. This property in fact is related to the convexity 2 of the risk surface to be minimized in the space of portfolios. Only if the surfaces are convex they will always be endowed with a unique absolute minimum and no fake local minima [10] and the risk minimization process will always pick-up a unique, well-diversified optimal solution. Therefore, though one can perfectly think of possible alternative axiomatic definitions of risk measure, we strongly believe that no sensible set of axioms could in any case admit sub-additivity violations. Constructing a Risk Measure In what follows we want to show how a coherent alternative to Value at Risk arises as the natural answer to simple questions on a specified sample of worst cases of a distribution. We will construct this measure in a bottom-up fashion to better appreciate that this construction does not leave much freedom and leads in a natural way to essentially one robust solution. For sake of concreteness, X will be the random variable describing the future value of the profit or loss of a portfolio on some fixed time horizon T from today and α = A% ∈ (0, 1) will be some percentage which represents a sample of "worst cases" for the portfolio that we want to analyze. Provided this information, the VaR of the portfolio with parameters T and A% is simply given by the loss associated with the related quantile x (α) of the distribution 3 . This statistic answers to the following question: What is the minimum loss incurred in the A% worst cases of our portfolio? Strange as it may sound, this is the most frequently asked question in financial risk management today. And due to that "minimum loss" in its definition VaR is not a sub-additive measure. Moreover, being simply the threshold of the possible A% losses, VaR is indifferent of how serious the losses beyond that threshold actually are. Little imagination is needed to invent portfolios with identical VaR and dramatically different levels of risk in the same A% worst cases sample. Any reader, at this point is tempted to modify the above question with the following: What is the expected loss incurred in the A% worst cases of our portfolio? We want to show that this is a good idea for at least two different reasons. First of all, because this question is undoubtedly a more natural question to raise when considering the risks of a specified sample of worst cases. Secondly, because it naturally leads to the definition of a sub-additive statistic as we will see in a few steps. It is not difficult to understand that if the distribution function of the portfolio is continuous, then the statistic which answers the above question is simply given by a conditional expected value below the quantile or "tail conditional expectation" [3]. For more general distributions however, this statistic does not fit question (4) since the event {X ≤ x (α) } may happen to have a probability larger than A% and is therefore larger than our 3 We will omit the T dependence where possible. set of selected worst cases. Indeed, T CE is a risk measure only when restricted to continuous distribution functions [9] while may violate sub-additivity on general distributions [2]. To understand which statistic is actually hidden in question (4) let us see how we would naturally answer to it having a large number n of realizations {X i } {i=1,...,n} of the random variable X. We simply have to sort the sample in increasing order and average the first A% values. To do this, define the order statistics X 1:n ≤ . . . ≤ X n:n as the sorted values of the n-tuple (X 1 , . . . , X n ) and approximate the number of A% elements in the sample by w = [nα] = max{m\|m ≤ nα, m ∈ N}, the integer part of n A%, a choice that for large n could be changed with any other integer rounding or truncation close to nα. The set of A% worst cases is therefore represented by the least w outcomes {X 1:n , . . . , X w:n }. Postponing the discussion of some subtleties on quantile estimation we can define the following natural estimator for the α-quantile x (α) . x (α) n (X) = X w:n (6) The natural estimator for the expected loss in the A% worst cases is then simply given by which we will call the A% Expected Shortfall of the sample. Note that then the natural estimator for T CE is in general an average of more than A% of the outcomes 4 . This may happen when the probability of the event X = x (α) is positive (the case of a discrete distribution function) so that there might be multiple occurrences of the value X i = X w:n . It is easy to see that ES (α) n is indeed sub-additive for any fixed n. Consider two variables X and Y and a number n of simultaneous realizations {(X i , Y i )} {i=1,...,n} . We can prove sub-additivity at a glance This result is very encouraging. If we understand which statistic ES (α) n is an estimator of for large n, we are likely to end up with a sub-additive measure. Notice, that a proof similar to (9) would fail for T CE (α) n . 4 We adopt the obvious notation 1 {Relation} = 1 , if Relation is true 0 , if Relation is false. Now, we can expand the definition of ES If we now had lim n→∞ X w:n = x (α) , with probability 1, it would be easy to conclude that with probability 1 we also have Well, this is the subtlety on quantile estimation we have mentioned. Equation (11) does not hold in general. Nevertheless it can be shown [2] that eq. (12) is more robust and in fact holds in full generality. We can then give the following Definition 2 (Expected Shortfall) Let X be the profit-loss of a portfolio on a specified time horizon T and let α = A% ∈ (0, 1) some specified probability level. The Expected A% Shortfall of the portfolio is then defined as This definition provides a risk measure perfectly satisfying all the axioms of definition 1. This explicit formulation was first introduced 5 in [1] where a general proof of sub-additivity 6 was also given which is not based on the n → ∞ limit of the above proof (9) of sub-additivity of ES (α) n . An implicit formulation of ES had already been proved to be coherent in [7], where however it was erroneously identified with T CE. Equation (13) might at a first glance look complicated. The concept it expresses is however simple as it is the literal mathematical translation of our above natural question and the limit for large n of the straightforward estimator (7). It is easy to realize that T CE, despite its simpler mathematical formulation (5) is on the contrary related to a much more complicated question than (4). To have a better insight of (13), the term x (α) (P[X ≤ x (α) ] − α) has to be interpreted as the exceeding part to be subtracted from the expected value E[X 1 {X≤x (α) } ] when {X ≤ x (α) } has probability larger than α = A%. When, on the contrary P [X ≤ x (α) ] = α, as is always the case if the probability distribution is continuous, the term vanishes and it is easy to see that (13) reduces to (5) or, in other words, ES (α) = T CE (α) . The actual simplicity of ES (α) can be appreciated only giving up defining it as a combination of expected values. There exists in fact an equivalent representation to (13) which reveals in a much more transparent way the direct dependence on the parameter α and on the distribution function F (x) = P [X ≤ x]. In fact, introducing the so-called generalized inverse function of one can easily show [2] that ES (α) can be simply expressed as the negative mean of F ← (p) on the confidence level interval p ∈ (0, α]: This is the most fundamental formulation of ES (α) . Its mathematical tractability makes it particularly appropriate for studying the analytical properties of ES (α) . For instance, continuity in α (which is a distinguishing property of ES (α) which T CE (α) and V aR (α) do not share) is manifest in (15) while it is not obvious in (13). An alternative useful expression equivalent to (13) has been recently formulated in [8] where the terminology 7 "α-Conditional Value at Risk" is however adopted for ES (α) with λ ≡ P[X ≤ x (α) ]/α ≥ 1. This relationship, which can be easily derived from (13) multiplying and dividing by P[X ≤ x (α) ], allows to put in evidence that in general ES (α) ≥ T CE (α) . Conclusions We started with an impasse coming from the fact that VaR was manifestly shown to be unfit for describing the risks of a portfolio and yet no valid practical alternative was still available in the class of eligible measures of risk. In this article we have seen that at least one specimen of the class of coherent risk measures allows us not to give up any of the advantages people got used to after the advent of VaR. ES is in fact universal: it can be applied to any instrument and to any underlying source of risk. ES is complete: it produces a unique global assessment for portfolios exposed to different sources of risk. ES is (even more than VaR) a simple concept since it is the answer to a natural and legitimate question on the risks run by a portfolio. Furthermore, any bank that has a VaR-based Risk Management system could switch to ES with virtually no additional computational effort. Even though a lot of work has still to be done to better investigate the statistical, probabilistic and computational issues raised by the use of ES, we believe that no serious difficulty will be encountered in adapting to it all the techniques developed in recent years for efficient calculations of VaR.	2014-10-01T00:00:00.000Z	2001-05-01T00:00:00.000	{ "year": 2001, "sha1": "33b28f72d5674237fd3bc003c5b3b60d463349a9", "oa_license": null, "oa_url": "http://arxiv.org/pdf/cond-mat/0105191", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "d08bac462b6d2b0178603eec6e14677a1fedb4c3", "s2fieldsofstudy": [ "Business", "Economics", "Mathematics" ], "extfieldsofstudy": [ "Economics", "Mathematics", "Physics" ] }
255260032	pes2o/s2orc	v3-fos-license	Hankel Norm Based Strict Passivity Performance of Digital Filter Using Saturation Nonlinearities The work proposes a new technique to examine and reduce the unusual memory effects in digital filters with saturation nonlinearity when external interference is present via strict passivity approach. Novel criterion is given to show the performance of Hankel norm in interfered digital filters by using strict passivity approach. The unusual behaviour of digital filters about previous excitations can be checked by using this proposed criterion. Given criterion ensures the asymptotic stability nature when there is no external input. The dominance of the proposed method is represented by a mathematical example. Introduction Since, last decades have shown the research importance in the field of signal processing digitally which leads to widespread investigation about the properties and behavior of digital filters [1]. Finite word length slows the performance while realizing the digital filters using any digital signal-based processors. Nonlinearities such as, quantization and overflow seem to happen when several signals are converted in discrete time systems using fixed point arithmetic, which may lead to limit cycles owing to self-sustained oscillations. Limit cycles results in instability of the filter [2][3][4][5][6][7]. In general, undesired response of systems results in ringing due to past excitations. Ringing seems to appear in several electronic systems, mainly due to noise signals (e.g., oscillatory nature in digital filters, parasitic inductance and capacitance in electronics circuits). Regular monitoring of the systems should be done in the system in order to clear the ringing effects as it may result in performance degradation or malfunction. In electro-mechanical system, sudden change in parasitic components to resonate creates ringing effect. In audio systems, too much feedback oscillation results in ringing. Ringing in the system usually stores the energy and yields memory effects which is undesired when applied external inputs. Hence, ringing can be computed based on performance of Hankel norm in the system. Therefore, Hankel norm performance quantifies the unwanted memory effects which appears in the past with external inputs on future outputs [24,25]. The basic concepts in the study of dynamical system are energy dissipation and consumption. Passivity symbolizes the energy consumption of a system and is used in various applications, e.g., electrical, chemical, mechanical, communication systems, etc. Passive systems are defined based on supply rate and storage function. The passivity theory provides a good method in analyzing the stability, design of complex energy-based systems [26,27]. Passivity is related to the property of input-output stability that is, if bounded input energy to the system yields bounded output energy. Passivity in state-space digital filters via saturation nonlinearity with external interference has been discussed in [19][20][21][22][23]. This work proposes a new method to quantify the unwanted memory effects in digital filters along with saturation nonlinearity via passivity approach based on the performance of Hankel norm. A new criterion based on performance of Hankel norm in digital filters is attained with saturation nonlinearity via passivity approach. Given proposed criterion in digital filters can examine and reduce the unwanted memory effects which appears in the past with external inputs on future outputs. Along with the presented criterion, improvement of the given result can be done by using the diagonally dominant matrix approach based on Hankel norm performance, which are given as remarks. The attained criterion ensures the stability as well as asymptotic stability when there is no external input in digital filter. The result given in this paper are novel, which is based on memory effect and passivity approach. The arrangement of the work is given as follows: Sect. 2 brought out the details about system taken for analysis trailed by criterion obtained for the stability of fixed-point digital filters with Hankel norm performance via passivity approach. Mathematical example is given in Sect. 3 to prove the dominance of the given proposed result. Section 4 presents conclusion of the paper. System Description for Hankel norm analysis The system under consideration of digital filter is where (k) ∈ n represents the state vector, (k) ∈ p denotes the linear combination vector of the states, (k) ∈ n indicates the external disturbance, a coefficient matrix S∈ n×n , and known matrix F ∈ p×n . For, saturation nonlinearities i ( i (k)) which are given as For a given > 0 and for time K > 0 the present work aims to find a stability criterion for the given system (1-3) with initial conditions as zero, when (r) ≠ . The digital filter has performance based on Hankel norm for if (5) is fulfilled. Condition given in (5) used to define effects which appears in the past with external inputs on the future output states for the system defined by (1)(2)(3). For the strict passive system, the condition follows as where is nonnegative constant and Ω( (k)) is a positive semi-definite storage function. Proof. By choosing a Lyapunov function of quadratic as. The first difference of (10) is specified by Then, for a positive scalar , we have Adding and subtracting T (k) (k) and by using (13), (11) can be rewritten as where (r) =2 T ( (k))L[ (k) − ( (k))] . It may be observed that (r) is nonnegative with respect to (4). Let us split the available storage function as ΔB(x(r)) and ΔB 1 (x(r)) which represents past and future output-based energy for the state B( (0)) [25]. Given LMI (8) is fulfilled, we have Sum on both the sides of (15) from 0 to K − 1 gives If LMI (8) is satisfied, we have Sum on both the sides of (15) from 0 to K gives Let us assume = B 1 ( (0)) . Then, Remark 1 Criterion in Theorem 1 is given in the form of Linear Matrix inequality (LMI) s and, hence, convex optimization procedures [28,29] are used to fix a feasible solution to the obtained LMIs. Remark 2 To reduce the potential conservatism of Theorem 1, the diagonally dominant matrix approach used in [11] can be applied. Remark 3 Theorem 1 ensures absence of overflow oscillations when interference is present in digital filters with Hankel norm performance via strict passive approach. Further, it also shows the asymptotic stability nature without external interference. Remark 4 In the realization of a digital filter, it is generally needed to choose a filter structure which does not exhibit limit cycles and delivers acceptable performance. Further, the work can be extended for multidimensional [30] systems which seems to be near future effort. Illustrative Examples Consider the subsequent example to confirm the superiority of the proposed result. Example 1 For simulation analysis, a digital filter (1)-(3) is considered with. With the assistance of LMI toolbox in MATLAB [19,20] one will be able to verify that given example for (8) and (9) has feasible values of unknown parameters. Conclusion The proposed work discusses the performance based on Hankel norm in digital filters via strict passivity approach. Novel LMI criterion describing the performance of Hankel norm in interfered digital filters via strict passivity approach are presented. By means of the obtained criterion in digital filters, it is easy to reduce the unwanted memory effects. The asymptotic stability property for digital filter when there is no external input are discussed. The worthful ness of the attained result is shown via mathematical example. Funding The authors declare that no funds, grants, or other support were received during the preparation of this manuscript. Data availability The data that support the findings of this study are openly available in reference [24].	2022-12-30T16:07:49.036Z	2022-12-28T00:00:00.000	{ "year": 2022, "sha1": "28cb281037ee3537277644eeacd44386fbab161f", "oa_license": "CCBY", "oa_url": "https://www.researchsquare.com/article/rs-1652577/latest.pdf", "oa_status": "GREEN", "pdf_src": "Springer", "pdf_hash": "498560ba7258691d3ad47b00a163b559202980b5", "s2fieldsofstudy": [ "Engineering", "Computer Science" ], "extfieldsofstudy": [ "Computer Science" ] }
102678151	pes2o/s2orc	v3-fos-license	Corrosion Behavior of L80Steel in Different Temperature and Sulfur Content To understand the corrosion behavior of L80 steel in different temperature and sulfur content, the experiment which simulated the downhole corrosive environment was conducted. From the experiment result, when other factors were constant, the lowest corrosion rate was appeared when the temperature was 90°C. The influence of sulfur was complex. When temperature was low, the corrosion rate was decreased with the increase of sulfur content and the experimental result was opposite when temperature was high. Introduction During the development of oil and gas reservoirs, there often exist acidic elements, such as CO 2 gas, H 2 S gas and sulfur element which may corrode the pipes. In recent years, as more and more high temperature and high acidic carbonate reservoirs have been, or will be, developed, it is increasingly imperative to address the pipe corrosions.L80 steel is the most common used steel in oil and gas fields. So it is meaningful to understand the corrosion behavior of L80 steel in different corrosive environment, especially in different temperature and sulfur content. According to the literatures presented before [1][2][3][4][5][6][7][8], most of the research focus on the influence of the partial pressure of sour gas, such as H 2 S and CO 2 . However in the production of oil field, sulfur which is the corrosive element was found in the content of oil. In this paper, the experiment which simulated the downhole corrosive environment was conducted to test the corrosion behavior of L80 steel. Experiment Preparation The sample of L80 was used for corrosion test. The chemical composition of L80 steel was shown in table 1.The size of the sample for corrosion test was 50×10×3mm. The sample treatment: First, the sample with size of 50×l0×3mm was cut from the block of the raw materials of L80 steel, polished by sand paper with grid degree of 300#-1000# and cleaned by alcohol and acetone, and dried for half of an hour in dryer. Finally the mass of the sample was weight out by scale. Experiment processing:①All of chemicals were added into the autoclave.②The support hold three parallel samples were put into the autoclave, and the autoclave was covered tightly.③The heater and stirrer were open to mix the medium and make the test temperature up to the need.④ The down-pressure value and flow counter were open, and the gas of H 2 S and CO 2 were inserted into the autoclave. The input value of the autoclave and down-pressure value of the vessel were shut down when the pressure of H 2 S and CO 2 was up to the design value. ⑤The corrosion test was started. The temperature fluctuation range was controlled to ±1°Cand the test time was continued to 720h during the test. ⑥After the experiment was finished. Firstly, the power of heater was shut off, and the sample was taken out from the autoclave after the autoclave was cooled to room temperature. Secondly, the corrosion production on the sample surface was cleaned according to the standard of "GB/16545-1996 corrosive product on the corrosive sample of metal and alloy", and the sample was cleaned and got rode of water using alcohol and acetone. Finally the mass of sample was weight and recorded after the sample cleaned was kept in the dryer for at least 24h. Experimental Result and Discussion The corrosion rate of L80 steel sample in corrosive medium was shown in table 3. The table 3 indicated that the lowest corrosion rate of 0.6833 mm·a -1 of the sample was arrived in the medium that was composed of Temperature of 90°C, S of 4% and water of 70%, and the highest corrosion rate of 2.0835 mm.a -1 of the sample was arrived in the medium that was composed of Temperature of 150°C, S of 4% and water of 30%. The effect of temperature, sulfur contenton the corrosion rate of the sample was shown as following: 1) Temperature: From the table 3 and figure 1, the lowest corrosion rate was achieved when the temperature is 90°C. It was seen that the corrosion rate of L80 steel decreased with the increasing of temperature firstly when the content of sulfur and water was constant. Then as the increase of temperature, the corrosion rate increased. Figure 1. The corrosion rate of L80 steel in different temperature 2)Sulfur: When sulfur content increased from 2% to 4%, and the content of temperature and water was constant, if the temperature of corrosion medium was lower than 90°C(60°Cand 90°C), the corrosion rate of L80 steel would decrease, and the corrosion would be inhibited. However if the temperature of corrosion medium was up to 150°C, the corrosion rate of L80 steel would increase with the increase of sulfur content.At the lower temperature the sulfur maybe only deposited on the surface of the steel. Although the deposition of sulfur could induce the severe localized corrosion, the general corrosion was mild. So the corrosion rate decreased with the increasing of sulfur content. Nevertheless at high temperature the sulfur would react with the corrosion product and the corrosion product would be destroyed. The protection and stabilization of corrosion film was destroyed. So the corrosion rate was increased with the increase of sulfur content. From the experimental result, the contribution of sulfur to the L80 steel can be summarized in two aspects as following: on the one hand, the deposition of sulfur could induce the severe localized corrosion; On the other hand, the sulfur could react with the corrosion product and corrosion film. Conclusions From the result of corrosion experiment, when other corrosive factors, such as water cut, sulfur content and the partial pressure of CO 2 and H 2 S gas were constant, the lowest corrosion rate was appeared when the temperature was 90°C. The influence of sulfur content was complex. When temperature was low(60°Cand 90°C), the corrosion rate was decreased with the increase of sulfur content. When temperature was high(150°C), the corrosion rate was increased with the increase of sulfur content.	2019-04-09T13:06:29.432Z	2017-10-01T00:00:00.000	{ "year": 2017, "sha1": "90307f55cb7481280020d872de2c48568ee69d05", "oa_license": null, "oa_url": "https://doi.org/10.1088/1757-899x/250/1/012058", "oa_status": "GOLD", "pdf_src": "IOP", "pdf_hash": "9855e7f7f99b101524326302d9091f491d26bd96", "s2fieldsofstudy": [ "Materials Science" ], "extfieldsofstudy": [ "Materials Science" ] }
253492275	pes2o/s2orc	v3-fos-license	Discovering and Targeting Dynamic Drugging Pockets of Oncogenic Proteins: The Role of Magnesium in Conformational Changes of the G12D Mutated Kirsten Rat Sarcoma-Guanosine Diphosphate Complex KRAS-G12D mutations are the one of most frequent oncogenic drivers in human cancers. Unfortunately, no therapeutic agent directly targeting KRAS-G12D has been clinically approved yet, with such mutated species remaining undrugged. Notably, cofactor Mg2+ is closely related to the function of small GTPases, but no investigation has been conducted yet on Mg2+ when associated with KRAS. Herein, through microsecond scale molecular dynamics simulations, we found that Mg2+ plays a crucial role in the conformational changes of the KRAS-GDP complex. We located two brand new druggable dynamic pockets exclusive to KRAS-G12D. Using the structural characteristics of these two dynamic pockets, we designed in silico the inhibitor DBD15-21-22, which can specifically and tightly target the KRAS-G12D-GDP-Mg2+ ternary complex. Overall, we provide two brand new druggable pockets located on KRAS-G12D and suitable strategies for its inhibition. In particular, several strategies have been proposed for targeting KRAS-G12D, such as indole-based inhibitors (BI-2852) of the Switch-I/II pockets [28,29], the piperazine-based compound TH-Z835 for ASP12 [30], the KRAS allosteric ligand KAL-21404358 targeting the site PRO110 [31], a multivalent small molecule named 3144 for pan-RAS inhibition [32] and a three cyclic peptide for GTP-bound KRAS-G12D [33]. It has only been in recent years that more works about G12D, G12R, G12S and G12V were reported and some specific inhibitors were proposed [34][35][36][37][38][39]. Despite these progresses, targeting the most prevalent and oncogenic KRAS-G12D mutation has remained elusive; no therapeutic agent has been clinically approved yet. Therefore, how to interrupt mutated KRAS-G12D oncogene remains a real challenge both in clinical and scientific research. Clinical data have implicated not only that mutated RAS isoforms vary by tissue and cancer types, but also within the most frequently KRAS-mutations, the mutated amino acid residues of KRAS are also different for each cancer. For instance, among KRAS mutation predominant cancers, KRAS-G12C mutations mainly occur in lung cancer (45%), whereas KRAS-G12D mutations are related to more than 50% of pancreatic ductal adenocarcinoma cases [40][41][42]. The KRAS mutations display high tissue-specific abilities to drive tumorigenesis, strongly suggesting that the mutation of only one KRAS residue can lead to significant mechanistic changes in KRAS behavior. This usually leads to consider that there are only minor differences at the molecular scale between different KRAS mutants, such as one-site GLY12 substitutions. Therefore, the detailed information on atomic interactions and local structures at the all-atom level of KRAS can be of crucial importance for oncogenic KRAS research [43]. Until now, several studies have been published on the effects of G12D mutation on the structure and conformation changes of KRAS [39,[44][45][46][47][48]. However, previous studies mainly focus on oncogenic proteins themselves and are less associated with the KRAS surroundings, although cancer is closely related to oncoproteins and the surrounding environment [49,50]. So, more detailed conformations and local structures of KRAS still remain to be revealed. As one of the important RAS cofactors, the ion Mg 2+ is coordinated in an octahedral arrangement with high affinity on RAS proteins [51][52][53]. Mg 2+ has been established essential for both guanine nucleotide binding and GTP-hydrolysis of RAS proteins. For instance, in HRAS, the difference in affinity between HRAS and guanine nucleotide in the presence or absence of Mg 2+ is ∼500-fold [54][55][56]. However, HRAS is rarely mutated in human cancers with only ∼10% rate found in bladder and cervical cancers [40][41][42]. The mechanism of Mg 2+ interaction with the most prevalent oncogenic KRAS species has not been investigated yet. In this paper, using microsecond scale molecular dynamics (MD) simulations at an all-atom level, we reveal that cofactor Mg 2+ plays a crucial role in the conformational changes of KRAS. The mutation of GLY12 site in KRAS-G12D triggers a distinct shift in the interaction patterns between Mg 2+ and KRAS, finally generating two druggable exclusive dynamic pockets on the KRAS-G12D-GDP-Mg 2+ ternary complex surface. In this paper, different from other works in the literature [46], we do not consider the association with the GTP species (which makes KRAS active) since it is related to the signaling of the oncogene, which can occur when attached to the cell membrane surface. Conversely, we consider the KRAS-GDP association (i.e., in the inactive state), which does not require the hugely computationally expensive incorporation of the cell membrane in an explicit way. With full use of the structural characteristics of these two dynamic pockets we designed in silico the specific inhibitor DBD15-21-22, a derivative of benzothiadiazine [57,58] (DBD), which can specifically and tightly target the KRAS-G12D oncogene, stabilizing the inactive state of KRAS-G12D-GDP. The reliability of this finding was validated by subsequent molecular dynamics simulations of KRAS-G12D together with DBD15-21-22, where, by analyzing the differences between the three-dimensional structure of DBD15-21-22 and guanine nucleotide and combining them with the molecular dynamics simulation results of DBD15-21-22 and wild-type KRAS, we can propose that DBD15-21-22 will be harmless for wild-type KRAS. Results and Discussion Firstly, we investigated the conformational changes of the three different isoforms of KRAS in aqueous ionic solution. The only difference between these three KRAS is located at codons 12, with the sequences and initial structures of the three isoforms shown in Figure A1 (Appendix A). Atomic detailed sketches of GDP and the main residues described in this part are reported in Figure A2. All meaningful atom-atom distances as a function of time and bond lifetimes are reported in Appendix A. The Fluctuations and Stability of KRAS Protein Conformations with Different Mutation Isoforms Root mean square deviations and root mean square fluctuations of KRAS-WT and the two mutant isoforms were firstly analyzed ( Figure A3A,B). Both properties are defined below. RMSD results show the fluctuations and stability of the conformations of the three species. An overall view of the evolution of conformational changes is shown in Figure A3C-E. We found that for KRAS-WT, there was a distinct conformational fluctuation around 1.9 µs during the simulation. In a similar fashion, for KRAS-G12C, a large conformation change was observed after 1 µs. KRAS-G12D exhibited a distinct conformational stability different from that of KRAS-WT and KRAS-G12C, with overall stability and no significant changes. From the perspective of residues, RMSF reveals its flexibility during the full simulation. Switch-I (SW-I) and Switch-II (SW-II) are the regions of the protein which were recently identified as potential drugging sites [42,48,59], and they show high conformational flexibility compared to other structures of KRAS. Noticeably, we observed that the flexibility of residues in the SW-I domain of KRAS-G12D is around 2-fold smaller than that of KRAS-WT and KRAS-G12C. Combining the RMSD and RMSF results, the KRAS-G12D demonstrated significant stability along full statistics 5 µs MD simulations (2.5 µs per trajectory). This suggests that (1) the different isoforms of KRAS are not sharing the same mechanisms in conformational changes; and (2) it corroborates that to a large extent, the conformational change of the KRAS protein is mainly embodied by SW-I and SW-II. These findings are in overall good qualitative agreement with previous computational studies [60]. Mutation of GLY12 Enhances the Interaction between P-Loop and SW-II Domain The GLY12 mutation of KRAS protein also shows significant effects on the interaction of P-loop with Switch-II ( Figures A4 and A5). GLY12 has weak hydrogen-bonding (HB) interactions with GLY60 and GLN61, but such HB interactions are enhanced when the GLY12 mutated to CYS or ASP. Compared with the GLY12 of KRAS-WT, residue CYS12 contains a sulfhydryl group such that the HB interaction between CYS12 (H1) and GLN61 (O1) is mildly enhanced (see Figure A4B). When GLY12 mutates to ASP, it contains two active HB electron donor sites (carboxyl group). This reverses the surface electrical properties of residue 12, with ASP12 having a longer side chain. The interaction between position 12 and residues GLY60 and GLN61 is significantly enhanced so that they can form three simultaneous strong HB with GLY60 and GLN61 ( Figure A5). In addition, the mutation of GLY12 also affects the interaction of VAL8 (near the P-loop) with THR58 (located on SW-II) ( Figure A6). In the KRAS-WT case, the HB VAL8(O3)−THR58(H2) and HB VAL8(O3)−THR58(H3) are alternatively generated, but the lifetime of HB VAL8(O3)−THR58(H2) is slightly longer than that of HB VAL8(O3)−THR58(H3) . In the KRAS-G12C case, the lifetime of HB VAL8(O3)−THR58(H2) is increased so that such a pair remains present for the entire simulation period, while HB VAL8(O3)−THR58(H3) interactions are mostly eliminated. Furthermore, in the case of KRAS-G12D, the lifetimes of both HB VAL8(O3)−THR58(H2) and HB VAL8(O3)−THR58(H3) are significantly enhanced, with these two HB being present throughout the whole simulation time. Notably, HB VAL8(O3)−THR58(H3) was more stable than HB VAL8(O3)−THR58(H2) in the KRAS-G12D case, which is a clear difference compared to KRAS-WT and KRAS-G12C. Strong Coordination Interactions between Mg 2+ and SER17, ASP57, GDP in KRAS-G12D Noncovalent interactions, such as hydrogen bonds and coordination bonds (CB), are important for proteins to maintain their tertiary structure as well as for protein-cofactor interactions. We analyzed the structure of the three different KRAS isoforms in the aqueous ionic solution by time-dependent atomic site-site distances between selected atomic sites to find out and estimate interactions between Mg 2+ and KRAS ( Figure A7). The full set of averaged values and lifetimes of HB/CB are also reported in Tables A2 and A3 of Appendix A. We selected in Figure 1 representative snapshots from the total computed length of 5 µs, where the pattern and mechanism of the formation and breaking of CB involving Mg 2+ are clearly seen. Interestingly, cofactor Mg 2+ exhibits a unique role in the conformational change of KRAS-G12D, while having little influence on KRAS-WT and KRAS-G12C. As shown in Figure 1, for the KRAS-G12D case, Mg 2+ can form CB with SER17 (oxygen labeled 'O2', see Figure A2), ASP57 (O1, O2) of KRAS-G12D and oxygens O1B and O2B of GDP (state 1). During the first 500 ns, CB Mg 2+ −SER17(O2) is in a dynamic fluctuation and can be formed and broken (state 2). In the next 1.5 µs, bonds CB Mg 2+ −GDP(O2B) and CB Mg 2+ −ASP57(O2) are broken. Finally, Mg 2+ forms strong CB with GDP (O1B), ASP57 (O1) and SER17 (O2) (state 3). In KRAS-WT and KRAS-G12C cases, the interaction patterns of Mg 2+ with KRAS and GDP are similar to each other, but they are very different from the KRAS-G12D case. In particular, Mg 2+ forms CB with SER17 (O2) and O2A, O1B, and O2B atoms of GDP, while the lifetimes of Mg 2+ -SER17 (O2) and Mg 2+ -GDP (O1B) are rather short. Once the CB Mg 2+ −SER17(O2) bond was broken, Mg 2+ was subsequently released from the binding site of KRAS-SER17. GDP Plays an Important Regulatory Role in the Conformational Change of SW-I In addition to the aforementioned CB/HB interactions in KRAS-GDP-Mg 2+ complex, the HB interactions were also investigated between SW-I and GDP ( Figure A8). We found that the HB interaction of SW-I (ASP30) with GDP is important for the transition of the SW-I conformation. The H2 and H3 atoms of GDP are very sensitive: when they form hydrogen bonds with the oxygen atoms of ASP30, these HBs between ASP30 and GDP can limit the conformational change of SW-I within a small range. From the time evolution of selected atom-atom distances d(t) in Figure A8, we can see that the strength of the HB interaction between ASP30 and GDP are KRAS-G12D > KRAS-WT > KRAS-G12C. This matches the RMSD variation of each KRAS in Figure A3A. Correspondingly, KRAS-G12D exhibits extreme conformational stability during intervals of at least 2.5 µs of the simulations. We also observed SW-I of KRAS-WT opening largely, while the HB interaction between ASP30 and GDP breaks around 1.9 µs. Finally, since ASP30 has the weakest HB interaction with GDP in the KRAS-G12C case, the latter exhibits large conformational changes around 1 µs. The Dominant Conformations of Wild-Type and Mutated KRAS Gibbs free energy profiles are of high significance to characterize the dominant conformations of KRAS during simulation. In the present work, such analytical tools will allow us to directly track the effects of GLY12 mutations on the dominant conformations of the KRAS. Since the main movement of the protein is concentrated in SW-I and SW-II and our main research object is KRAS-GDP-Mg 2+ ternary complex, we chose to compute Gibbs free energy landscapes by using two specific variables, including RMSD and radius of gyration, see below. The method employed to obtain the free energy profiles is the so-called "principal component analysis" [61,62], where the components RMSD and R g work as reaction coordinates ( Figure 2). In the KRAS-WT case, we detected two free energy basins (I, II). Basin II is the one with lowest free energy (set to 0 kJ/mol), and it was chosen as the reference so that basin I shows a barrier of 4.6 kJ/mol, when transitions from basins I to II are considered. These two Gibbs free energy basins represent the two main stable states of KRAS-WT during MD simulations. We can obtain from the MD trajectories that in state I, SW-I is slightly open, while in the dominant basin II, SW-I opens largely and SW-II tends to be closer to the P-loop (region containing the mutated codon 12 and depicted in orange color). Several Gibbs free energy basins were also detected in the KRAS-G12C. In this case, the representative conformation of basin I is similar to the corresponding one in KRAS-WT, while basin II is divided into several small energy basins, with IIa and IIb as examples. The dominant conformation is now located on basin IIb (0 kJ/mol), with basin I at 2.7 kJ/mol and basin IIa at 1.0 kJ/mol. These basins are separated by low free energy barriers and they are moderately easy to access by the system. The differences between basins IIa and IIb are mainly on SW-I, with the SW-I conformation of IIb being more open than that of IIa. The comparison between KRAS-WT, KRAS-G12C and KRAS-G12D shows a clear difference: a single Gibbs free energy basin is found for KRAS-G12D, which indicates that such species have only one dominant stable conformation. As it can be seen from the representative crystal structure of this particular free energy basin I ( Figure 2C), this dominant conformation of KRAS-G12D corresponds to SW-I being slightly open, while the distance between SW-II and P-loop is relatively tight. This is because the strong CB interaction of Mg 2+ with GDP (O1B), SER17 (O2) and ASP57 (O1) can stabilize the distance between SW-II, GDP and the α-helix of KRAS, where SER17 is located. We also observed that the ASP12 shows strong interaction with GLY60 and GLN61, both located at SW-II. At the same time, hydrogen atoms H2 and H3 of GDP can form stable HB interactions with ASP30, helping to stabilize the distance between SW-I and GDP. To corroborate the full convergence of the free energy landscapes reported in Figure 2, we report in Appendix A ( Figure A9) the contribution of the two independent 2.5 µs trajectories that we employed in this work (A,B), compared with their average (C), as shown in Figure 2C. The Two Unique Druggable Dynamic Pockets on KRAS-G12D From the previous analysis, KRAS-G12D exhibited a stable conformation that was significantly different from those of KRAS-WT and KRAS-G12C, and Mg 2+ exhibited a unique binding mode in the KRAS-G12D surface. Mg 2+ can form strong CB interactions with GDP (O1B), SER17 (O2) and ASP57 (O1). Well-established experimental results [51,52] showed that the octahedral structure is the most stable structure for Mg 2+ first shell coordination, which is fully consistent with our simulation results. When we take into account the interaction of KRAS-GDP-Mg 2+ complexes in aqueous Mg-Cl solution, we find the dynamic water pocket I in the Mg 2+ binding area (left side of Figure 3). We call it "dynamic" in the following sense: in the octahedral structure of Mg 2+ , the fluctuation frequencies of the coordinated water molecules (in positions H 2 O-1, H 2 O-2 and H 2 O-3) are lower than those of free water molecules in solution, i.e., the former remain in their positions for much longer periods of time than the latter (nanoseconds compared to the picosecond time scale for bulk water HB dynamics [63]). At the same time, the coordinated water molecules can be exchanged with other water molecules in the solution ( Figure A10). Further analysis of the trajectory uncovered another dynamic water pocket II, which is located between the phosphate group of GDP and the ASP12 (right side of Figure 3). This dynamic pocket can accommodate only one H 2 O molecule, through HB interactions of H 2 O with GDP (O2B) and ASP12 (O1/O2) ( Figure A11). Further, the water molecule in this pocket is in a dynamic fast exchange with other water molecules in the solution. For the sake of corroboration of the newly observed pockets I and II, we included in Appendix A (see Figure A12) a comparison of the location of the pocket reported by Kessler et al. [28] with the pockets reported in the present work. Structure-Based Drug Design and Targeting the Dynamic Water Pockets on KRAS-G12D In order to target the newly revealed binding pockets, we designed a drug in silico able to block GDP at its main cavity in a permanent way. There are several identified ways to target KRAS, the following being the most relevant [64]: inhibition of RAS expression, interference with RAS post-translational modifications, inhibition of RAS function or targeting specific downstream effectors. Furthermore, our strategy is different from that of the designers of recent drug Sotorasib [65]. We observed that in recent years, selective binding ligands for Mg 2+ have gradually become known [66]. In dynamic water pocket I, H 2 O-2 and H 2 O-3 molecules can exchange with other water molecules in the aqueous solution. The two oxygen atoms of H 2 O-2 and H 2 O-3 are on the same horizontal plane, and the distance between the two oxygen atoms is comparable to the size of the bidentate ligand. This feature makes it possible to design the selective inhibitors that can replace these two water molecules and target this dynamic water pocket I. Herein, we designed a series of inhibitors for these two dynamic water pockets using a benzothiadiazine, specifically 3,4dihydro-1,2,4-benzothiadiazine-1,1-dioxide (C 7 H 8 N 2 O 2 S, DBD), as a template. The most suitable species obtained is designed as DBD15-21-22 (see Figure 4A) since it is a DBD derivative, similar to some of those reported in Ref. [67]. DBD15-21-22 Can Target KRAS-G12D and Bind Dynamic Water Pockets I and II To further study the structure of the DBD15-21-22 interaction with KRAS-G12D, we performed microsecond time scale MD simulations of DBD15-21-22 together with KRAS-G12D. As we can see from the MD results, the DBD15-21-22 is tightly bound to its binding pocket ( Figure 4B). The N4 and N5 of DBD15-21-22 can replace H 2 O-2 and H 2 O-3 in the octahedral structure of Mg 2+ and then specifically bind to the dynamic water pocket I. Benefiting from the location and size of the amino group (−NH 2 ) near the N4 and N5, the −NH 2 group can perfectly replace the H 2 O-4 in the dynamic water pocket II. These two concerted actions enable DBD15-21-22 to bind tightly and specifically to the two dynamic water pockets on the KRAS-G12D-GDP-Mg 2+ complex. In addition, other active sites in DBD15-21-22 can also form stable HB interactions with other residues of KRAS-G12D. In particular (see Figure 4A), the H1 atom of DBD15-21-22 can form HB with ASP57 (O2) and the H11 atom on the phenolic hydroxyl (−OH) group can form HB with multiple residues on SW-I to lock the conformational change of SW-I. In order to estimate the detailed HB and CB interactions between DBD15-21-22 and its unique druggable pocket on KRAS-G12D, we display the time evolution of selected atom-atom distances d(t) in Figures A12-A18. DBD15-21-22 Is Harmless to KRAS-WT As we can see in Figure 5A, in the structure of DBD15-21-22 there exists a guanylate moiety but the binding patterns are quite different from GDP/GTP. When GDP/GTP binds to the GDP/GTP binding pocket ( Figure 5E), the O3 of guanine goes deep inside the pocket. DBD15-21-22 also has a guanine group, which is similar to GDP/GTP. The difference is that in DBD15-21-22 ( Figure 5A), the DBD part is connected to the N3 atom of guanine; in contrast, the β-D-ribofuranosyl in the GDP structure is connected to the N4 atom of guanine ( Figure 5C). The drug we designed takes advantage of this subtle difference, and prevents the guanine structure in the DBD15-21-22 from binding to the GDP/GTP pocket through the steric hindrance of the DBD part in the DBD15-21-22. Moreover, we investigated the three-dimensional structure of DBD15-21-22 and GDP in an aqueous solution ( Figure 5D To further verify the low toxicity of DBD15-21-22 to KRAS-WT, we also performed MD simulations of DBD15-21-22 together with GDP free KRAS-WT. From the simulation results, we can see that DBD15-21-22 was unable to bind to the GDP/GTP binding pocket on KRAS-WT but was dissolved in an aqueous solution most of the time ( Figure A19A).This finding is further verified by the radial distribution function (RDF) of the DBD15-21-22 interaction with aqueous solutions ( Figure A19B). We can observe that DBD15-21-22 dissolves well in water, with hydrogen H1 and nitrogen N5 of DBD15-21-22 both forming HBs with H 2 O. However, in the KRAS-G12D-DBD15-21-22 simulation case, the HB interaction of DBD15-21-22 with water was interrupted such that DBD15-21-22 was well bound to the druggable pocket on the surface of KRAS-G12D, as described in Figure 4B). Methods and Materials Our main computational tool has been microsecond scale molecular dynamics. In MD, after the choice of reliable force fields, the corresponding Newton's equations of motion are integrated numerically [68], allowing us to monitor each individual atom in the system in a wide variety of setups, including liquids at interfaces in solid walls or biological membranes, among others [69,70]. We fixed the number of particles, the pressure and the temperature of the system, while the volume was adjusted accordingly. MD can model hydrogens at the classical [71,72] or quantum levels [73] and, in addition to energetic and structural properties, MD provides access to time-dependent quantities, such as the diffusion coefficients, rotational and vibrational motions [74] or spectral densities [75], enhancing its applicability. In the present work, we conducted MD simulations of three KRAS isoforms with sequences represented in Figure A1. Each system contains one isoform of KRAS-GDP complex fully solvated by 5697 TIP3P water molecules [76] in potassium chloride solution at the human body concentration (0.15 M) and magnesium chloride solution concentration (0.03 M), yielding a system size of 19,900 atoms. All MD inputs were generated using the CHARMM-GUI solution builder [77][78][79], and the CHARMM36m force field [80] was adopted for KRAS-GDP-Mg 2+ interactions. The force field used also includes the parameterization of the species GDP (it can be searched as "GDP" in the corresponding CHARMM36m topology file: https://www.charmm-gui.org/?doc=archive&lib=csml (accessed on 10 October 2021)). All bonds involving hydrogens were set to fixed lengths, allowing fluctuations of bond distances and angles for the remaining atoms. The crystal structure of GDP-bound KRAS proteins was downloaded from the RCSB PDB Protein Data Bank [81], file name "4obe". The three sets of KRAS-GDP complex (wild type, G12C mutate and G12D mutate) were solvated in a water box, all systems were energy minimized for 50,000 steps and well equilibrated (NVT ensemble) for 250 ps before generating the production MD. Two independent production runs were performed within the NPT ensemble for 2.5 µs. All meaningful properties were either averaged from the two runs.The pressure and temperature were set at 1 atm and 310.15 K respectively, in order to simulate the human environment. In all MD simulations, the GROMACS/2021 package was employed [82]. Time steps of 2 fs were used in all production simulations, and the particle mesh Ewald method with a Coulomb radius of 1.2 nm was employed to compute long-ranged electrostatic interactions. The cutoff for Lennard-Jones interactions was set to 1.2 nm. Pressure was controlled by a Parrinello-Rahman piston with damping coefficient of 5 ps −1 , whereas temperature was controlled by a Nosé-Hoover thermostat with a damping coefficient of 1 ps −1 . Periodic boundary conditions in three directions of space were taken. We employed the "gmx-sham" tool of the GROMACS/2021 package to perform the Gibbs free energy landscape analysis. Other alternatives, such as transition path sampling [83][84][85] or metadynamics [86,87], were not considered here because of their high computational cost for large systems. The designed inhibitor DBD15-21-22 was parameterized using CgenFF [88,89] to obtain the CHARMM-compatible topology and parameter files. Further, we switched to run another four 1 µs of MD simulations to study the interaction of the inhibitor DBD15-21-22 with KRAS-G12D and to verify that DBD-15-21-22 is harmless on the wild-type KRAS. These four new MD simulations have the same setup as the previous ones. Moreover, the software VMD [90] and UCSF Chimera [91] were used for trajectory analysis and visualization. As key computed quantities in this work, the radius of gyration (R g ) is defined in Equation (1): where m i is the mass of atom i, and r i the position of the same atom with respect to the center of mass of the selected group. Root mean square deviations (RMSD) are defined by Equation (2): where δ i is the difference in distance between the atom i (located at x i (t)) of the catalytic domain and the equivalent location in the crystal structure, and finally root mean square fluctuations (RMSF) are defined by Equation (3): wherex i is the time average of x i , and ∆t is the time interval where the average is taken. Conclusions To summarize the main findings of the work, let us remark that KRAS mutations are widely present in many cancer cases, but there exist usually only molecular-scale differences between different KRAS mutants, such as in the one site GLY12 mutation. This means the direct information on atomic interactions and local structures at the all-atom level of KRAS can be extremely useful for oncogenic KRAS research. Based on powerful computer simulation tools, we focused our analysis on the local structure of KRAS proteins when associated with Mg 2+ , GDP, and water. After systematic analysis of meaningful equilibrated data, we can reveal the dynamic differences between KRAS-WT and the mutated species KRAS-G12C and KRAS-G12D at the all-atom level. There exist several important intermolecular interactions affecting the behavior of KRAS. The most relevant features are (1) the strong coordination interactions between Mg 2+ , SER17, ASP57 and GDP in KRAS-G12D; (2) the mutation of GLY12 enhances the interaction between P-loop and SW-II domain, with the enhancement magnitude ordered as KRAS-G12D > KRAS-G12C; and (3) the GLY12 mutation also affects the interaction between GDP and SW-I. Although GLY12 mutation can enhance the interaction of the P-loop with SW-II, in the KRAS-G12C case, the interaction of GDP with SW-I was almost totally eliminated, whereas the G12D mutation greatly enhanced the interaction of GDP with SW-I. In Figure 6, we show the alignment of three representative dominant structures of the GDP-bound KRAS-WT, KRAS-G12C and KRAS-G12D. As shown in Figure 6A, the downward sliding of the α-helix and β-sheet structures of KRAS-WT promoted the large opening of the SW-I domain. When GLY12 was mutated to ASP12, the interaction between P-loop and SW-II was largely enhanced, then the relative position of P-loop and SW-II was shortened and stabilized. This promotes strong coordination interactions between Mg 2+ , SER17, ASP57 and GDP, which further stabilizes the relative positions of β-Sheet and SW-II in the protein structure. Also benefiting from the hydrogen bond interaction between ASP30 and GDP, the fluctuation of SW-I was limited to a small range, which led to a slightly open stable conformation of SW-I in KRAS-G12D. When GLY12 was mutated to CYS12, although the interaction between P-loop and SW-II was enhanced, it was not enough to stabilize the relative positions of α-Helix, β-Sheet and SW-II in the structure of KRAS, and the conformational changes were similar to those of the KRAS-WT. After deciphering the laws of conformational changes of different KRAS mutants, two druggable dynamic water pockets on KRAS-G12D-GDP-Mg 2+ ternary complex were revealed. We designed a specific inhibitor DBD15-21-22 for these dynamic water pockets, with the aim to lock the KRAS-G12D in the "off" (inactive) state, i.e., when the protein is unable to perform signaling processes and eventually develops cancer. Benefiting from the fact that H1 and O3 atoms of DBD15-21-22 can form intramolecular hydrogen bonds in aqueous solution, we detected no binding effects of DBD15-21-22 toward the GDP/GTP binding pocket of KRAS-WT, whereas our molecular dynamics simulation results show that this KRAS-G12D inhibitor can efficiently bind to the targeting pocket located between SW-I, SW-II and P-loop, thereby efficiently locking KRAS-G12D in the "off" state. Overall, our work provides a new druggable pocket and a reliable protocol for the development of specific inhibitors targeting KRAS-G12D. The three classes of KRAS proteins that were considered in this work are reported in Figure A1, where the initial structure of KRAS-WT, was acquired from PDB bank (4obe.pdb). We also show the octahedral structure of Mg 2+ in the crystal structure, consistent with relevant experimental studies [51,52]. In order to make clear the atomic sites that will be described and analyzed in the article, we represent sketches of GDP and the main amino acid structures analyzed in the present work in Figure A2. In Figure A3, and from the RMSD values as a function of time, we can see that the conformation of KRAS-G12D is stable during the whole simulation process (total statistics 5 µs). The difference from KRAS-G12D is that the conformation of KRAS-WT protein changes greatly in about 1.9 µs, and the conformation of KRAS-G12C protein changes greatly in about 1 µs. The RMSF results show the flexibility of amino acid residues in the KRAS structure. From the RMSF results, it can be seen that the structurally active regions of KRAS protein are mainly Switch-I (SW-I) and Switch-II (SW-II) domains. Notably, the flexibility of amino acid residues in the SW-I domain of KRAS-G12D is around twofold smaller than that of KRAS-WT and KRAS-G12C. Appendix A.2. Atomic Distances and KRAS Conformations Throughout the work, it is very important to make sure that binding connections between specific atomic sites are stable enough to be considered as meaningful interactions so that monitoring the instantaneous distances between tagged species is of crucial importance. In our case, the computation of radial distribution functions is not suitable in most cases, given the specificity of the atomic sites involved in the relevant bindings between ions, amino acids and drugs. The series of distances are reported in Figures A4-A8, A10, A11 and A13-A18, together with snapshots corresponding to the configurations of highest biophysical relevance. We mainly monitor hydrogens bonds and coordination bonds. For instance, in Figure A6, we compare the effect of different mutations on the interactions between amino acid residues (VAL8-THR58) in the KRAS protein. In KRAS-WT, the HB interaction can last ∼2 µs. When the GLY12 mutated to CYS, the HB interaction between VAL8 and THR58 have been enhanced, interestingly, VAL8 (O3) prefers to form hydrogen bonds with THR58 (H2). In KRAS-G12D case, the hydrogen bond interaction between VAL8 and THR58 is significantly enhanced compared to KRAS-WT and KRAS-G12C. In Figure A7, the coordination bond analysis of Mg 2+ is shown. In KRAS-WT case, Mg 2+ cannot form a CB with ASP57; in the first ∼300 ns, Mg 2+ can form CB with SER17 (O2) and GDP (O2A,O1B,O2B) and the CB Mg 2+ −SER17(O2) and CB Mg 2+ −GDP(O1B) are broken, while the CB Mg 2+ −GDP(O2A) and CB Mg 2+ −GDP(O2B) are very stable. In KRAS-G12C case, the situation is similar to KRAS-WT. The only difference is that the coordination interaction between Mg 2+ and SER17 (O2)/GDP (O1B) can last ∼400 ns. Finally, in the KRAS-G12D case, the difference are that Mg 2+ can form stable CB interaction with ASP57, SER17 and GDP (O1B). In the first ∼250 ns, the coordination interactions between Mg 2+ , SER17, ASP57 and GDP fluctuated, and then Mg 2+ formed stable coordination bonds with SER17 (O2), ASP57 (O1) and GDP (O1B). From the selected atom-atom distance data of Figure A8, we can see that the structure of KRAS-G12C is similar to that of KRAS-WT, with the hydrogen bond interaction between GDP and ASP30 being very weak. However, when GLY12 is mutated to ASP12, the interaction between GDP and ASP30 has been significantly enhanced. The dynamic change law of three water molecules (in positions H 2 O-1, H 2 O-2 and H 2 O-3) in the dynamic water pocket I is similar, so in Figure A10 we take the dynamic change law of water molecules at position H 2 O-1 as an example. From the selected atomatom distance data, we can see that the exchange period between water molecules in dynamic water pocket I and solution water molecules is ∼1.7 µs. The dynamic water pocket II is consists of the oxygen atom (O2B) of GDP and the oxygen atoms of ASP12 (O1,O2), as shown in Figure A11 where in left side we show the representative snapshot of dynamic water pocket II and in the right side the distance between ASP12 (O1,O2) and GDP (O2B). As shown in the left side of Figure A11, the dynamic water pocket II can accommodate one water molecule. And this water molecular has a high exchange frequency with the water molecule in the aqueous solution, with an exchange period of about ∼100 ps. Therefore, the time evolution of atom-atom distance d(t) corresponding to the water molecule and GDP/ASP12 is not calculated. In addition to the dynamic pockets I and II detailed above, the targetable pockets reported by Kessler et al. [28] were also monitored in our KRAS-G12D case. We show in Figure A12 a comparison of the main findings of the present work compared to those of Kessler et al. The difference in the location and characteristics of the two pockets is made clear. The dynamics of the designed drug DBD15-21-22 (H1) indicates ( Figure A15B) that at the beginning of the trajectory it is very far from the ASP57(O2) but after 50 ns it becomes close to the ASP57 (O2) and capable to form stable hydrogen bonds interactions. Finally, Figure A19, shows selected radial distribution functions of DBD15-21-22 when dissolved with the KRAS-WT species compared to the KRAS-G12D-GDP case: in the former WT case, the inhibitor DBD15-21-22 is solvated by water instead of remaining at the surface of KRAS. Appendix A.3. Convergence of the Simulations The equilibration of our simulation systems is reached after a few tens of nanoseconds. Afterwards, two long independent trajectories of 2.5 µs each are produced. In order to show the convergence of our results, we report the calculation of the free-energy surface corresponding to the KRAS-G12D mutation case from the two runs (A, B) and its average, (C) as shown in Figure A9. We include a superposition (D) of the stable state configuration of KRAS-G12D for the three sets reported (A, B, C). Figure A9. The system including KRAS-G12D reveals that the two generated trajectories of 2.5 µs produce a very similar free-energy surface (cases A,B) with the single stable state located approximately at the same coordinates. The average (C) is reported in Figure 3 of the manuscript. Stable states indicated by IA, IB and IC. Snapshot (D) reveals small differences for the three configurations corresponding to the stable state. Note: here, we choose SW-I, SW-II, GDP and Mg 2+ to calculate RMSD and R g , so the RMSD value here is larger than in Figure A3A. In Figure A3A, the RMSD is calculated for the full KRAS protein. Table A2. Estimated lifetimes (in ns) of HB and CB interactions (Section 2 of main text, corresponding to Figures A4-A8). Total simulation time of 5 µs (2.5 µs per trajectory). We indicate in "Bond" the two amino acids or ion involved in the HB or CB and the corresponding atoms (hydrogen, oxygen) in parentheses. Double labels indicate averaged values.	2022-11-13T16:31:24.273Z	2022-11-01T00:00:00.000	{ "year": 2022, "sha1": "88190e54c62623c5cc659cf290ed37308bf8eccb", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1422-0067/23/22/13865/pdf?version=1668081325", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "7e1bed4d7d3196b2c7f339d1afa1244bbc8bfecf", "s2fieldsofstudy": [ "Biology", "Chemistry" ], "extfieldsofstudy": [ "Medicine" ] }
225442990	pes2o/s2orc	v3-fos-license	Enabling IoT Network Slicing with Network Function Virtualization Numerous Internet of Things (IoT) devices are being connected to the networks to offer services. To cope with a large diversity and number of IoT services, operators must meet those needs with a more flexible and efficient network architecture. Network slicing in 5G promises a feasible solution for this issue with network virtualization and programmability enabled by NFV (Net-work Functions Virtualization). In this research, we use virtualized IoT platforms as the Virtual Network Functions (VNFs) and customize network slices enabled by NFV with different QoS to support various kinds of IoT services for their best performance. We construct three different slicing systems including: 1) a single slice system, 2) a multiple customized slices system and 3) a single but scalable network slice system to support IoT services. Our objective is to compare and evaluate these three systems in terms of their throughput, average response time and CPU utilization in order to identify the best system design. Validated with our experiments, the performance of the multiple slicing system is better than those of the single slice systems whether it is equipped with scalability or not. IoT network slicing is enabled by NFV based on the MANO framework. We run virtualized IoT platforms as our VNFs and customize network slices through our NSD (Network Service Descriptor) to support IoT services of various QoS. We propose to customize each network slice first with different bandwidth to handle different types of IoT services. In addition, because of the virtualization of IoT platforms, we can scale out/in their instances rapidly and dynamically to support the variation in service load [8]. Hence three different slicing systems are constructed for our research. The first system consists of only a single network slice for all IoT services, while the second one consists of three customized slices to handle each type of IoT traffic separately. The last system is similar to the first one but can scale out/in VNFs on the slice. To evaluate the performance of each system, we design a Traffic Generator to simulate three types of IoT services with different bandwidth requirements. We show the advantages and disadvantages of each system and articulate performance tradeoffs by analyzing their pros and cons. The rest of the paper is organized as follows: Section 2 introduces the background information of oneM2M, ETSI NFV architectural framework and network slice. Section 3 presents our system design and system workflow in Open-Stack. Section 4 describes three different systems and compares their performance evaluation. Section 5 presents our system design, implementation and evaluation in Kubernetes. Finally, Section 6 shows the conclusion and future work of this paper. Background In this section, we explain the oneM2M IoT platform we used in our system, the NFV architectural framework and the concept of network slicing. ETSI NFV Architectural Framework NFV MANO (Management and Orchestration) [11] is a framework developed by ETSI (Europe Telecommunication Standards Institute) for the management and orchestration of all virtualized resources including compute, network, storage, It is also responsible for managing the life cycle of VNFs, such as instantiation, scaling, update and termination. In addition, it manages the policy of network services, the collection and transfer of performance measurement, and the allocation of resources related to infrastructures. The NFV MANO framework is adopted in our research to construct the network slicing environment. As illustrated in Figure 2, NFV MANO consists of three major components. • NFV Orchestrator (NFVO), which is in charge of the lifecycle of Network Services (NS) and responsible for onboarding Network Service Descriptor (NSD). • VNF Manager (VNFM), which is responsible for the lifecycle management of the VNFs including VNF scaling out/in and their performance and fault management. Network Slice According to 3GPP [1], network slice is an end-to-end network architecture. It consists of multiple network slice subnets. Each network slice subnet represents different components in an end-to-end network such as access network, core network, transport network. NFV MANO, as the key enabling technology of network slicing, maps each network slice subnet to an NS in MANO. Each NS is defined by an NSD that consists of a set of attributes and several constituent descriptors including VNF descriptors (VNFDs), VL descriptors (VLDs), and VNFFG descriptors (VNFFGDs) [12]. The attributes of NSD are used to specify how NS instances should be deployed. Network slicing enables the operator to divide a physical network into multiple virtual and logically independent end-to-end networks. Each network slice is tailored to fulfill different service requirements, such as delay, bandwidth, security, and reliability to cope with diverse network application scenarios [13]. Mobile operators can use network slicing to provide customized 5G networks to various vertical services based on specific needs of each [14]. • OpenStack is an open source cloud operating system for virtualizing and managing resources including compute, network and storage. It provides multiple managing services such as Nova for compute, Neutron for network [20], Cinder for storage, Keystone for key management, Horizon for dashboard and Heat for orchestration. There are other orchestrators that can be used as NFVO and VNFM, such as Open Network Automation Platform (ONAP) [21], Open Source MANO (OSM) [22] and Open Baton [23]. Since Tacker is an official OpenStack project, compared to other open sources, it is highly compatible with OpenStack VIM. Moreover, its design has the advantage of simplicity that allows users to easily deploy and operate. Therefore, we adopt Tacker as NFVO and VNFM in the NFV MANO framework. In our system, OM2M IN instances are deployed as the VNFs. On the other hand, NSs are the composition of OM2M IN instances and a Load Balancer to be introduced next. We construct the three slicing systems for our research experiments to support IoT services including: 1) a single slice system, 2) a multiple customized slices system and 3) a single but scalable network slice system. Our objective is to compare and evaluate these three systems in terms of their throughput, average response time and CPU utilization in order to identify the best system design. System Architecture We first explain the functional blocks of our three systems, then show the Network Service lifecycle management flows and the system workflows. The general architecture of our systems is illustrated in Figure 3 and our three systems are depicted in Figure 4. Note that we design three new system components Master Node, Load Balancer and Traffic Generator on top of OpenStack and Tacker open sources in order to complete our systems. • Master Node is incorporated in VNFM to monitor the CPU status of VNFs on each network slice in order to trigger scale-out or scale-in actions [24]. When the average CPU usage of VNFs exceeds an overload threshold, it will trigger the scale-out of the VNFs on the network slice. On the other hand, if the CPU usage is lower than an underload threshold, it will trigger the scale-in of the VNFs. • Load Balancer is designed to fairly dispatch the incoming traffic to each VNF [25]. It is used only in the single slice scalable system (see Figure 4(c)) for distributing traffic. We use RabbitMQ [26], an open-source message-broker software implementing Advanced Message Queuing Protocol (AMQP) [27], based on Remote Procedure Call (RPC) to design our load balancer. In our system, Traffic Generator sends HTTP requests to Load Balancer and Load Balancer sends those requests to a load balancing queue. On the other side, each OM2M VNF as a server consumes those requests from the queue and replies a response back to Load Balancer. • Traffic Generator is a multi-thread program that we design to simulate three types of IoT traffic. It can set the number for each kind of ASN devices and the frequency of sending data. Three types of IoT traffic generated include video, adaptive lighting and smart parking. Each traffic is a stream of HTTP requests. 1) Video: This is to simulate a security surveillance service enabled by the video camera. It provides monitoring services for road traffic and crowd movement. This service has the highest bandwidth demand among all three types of traffic. 2) Adaptive lighting: This is to simulate an adaptive lighting service by the smart street light pole that monitors weather conditions and adapts the brightness of street lighting accordingly based on the inputs from temperature, humidity, air pollution and light sensors. 3) Smart parking: This is to simulate a smart parking service that monitors the availability of parking spaces based on geomagnetic sensors embedded in parking areas. This service has the lowest bandwidth requirement among these three types of IoT services. Figure 4 shows the differences of network slices in three systems under study. In our experiment, Traffic Generator will simulate the same three types of traffic. But the three systems would handle traffic in different ways. As depicted in Figure 4(a), the single slice system architecture has only one network slice with one IoT platform as a VNF. The only one VNF needs to handle all types of IoT traffic. Then Figure 4(b) shows the multiple slicing system architecture that has three types of network slices. Each network slice is provisioned with a different customized bandwidth for dealing with a particular type of IoT service. At last, Figure 4(c) shows the design of the single slice scalable system architecture where only one network slice is provisioned but this slice supports scalability. Similar to the single slice system, its network slice needs to handle all three types of IoT traffic. But unlike the single slice system, it is capable of scaling its VNF to multiple instances and thus is equipped with a Load Balancer to evenly distribute IoT traffic to multiple VNF instances. However, because of Load Balancer, IoT traffic must go through an additional VNF, which may result in a longer response time than other systems. There are four phases to run a network slicing system. Network Service Lifecycle Management Flows In the preparation phase, we set up the environment by first registering an OpenStack VIM to the Tacker System Workflow for Scaling The workflow of our system for scaling is shown in Figure 6. This scaling mechanism is only used by the single slice scalable system in our experiment. Note scale-in threshold, it will trigger the scale-in action. However, if there is only one VNF left for the NS, the scale-in action will not be triggered. In addition, if there is already an action being executed, the next scale-out or scale-in action will not be triggered until the previous one ends. Implementation and Evaluation in OpenStack In this section, we show our test environment setup and experimental results. Three types of traffic are simulated through Traffic Generator designed to evaluate the performance of each system. The evaluation metrics include throughput, average response time and CPU utilization. Test Environment Setup Our test environment consists of two servers. Both Tacker and OpenStack are running on these two servers configured as shown in Table 1. Table 2 shows the virtual resource allocation of each VNF in our environment. Experimental Results In our experiment, we use Traffic Generator to simulate three types of traffic Table 3. The required bandwidth is set to twice the expected traffic throughput to avoid temporary excessive traffic. For the single slice system and the single slice scalable system, we set the bandwidth to 1400 Kbps. For the multiple slicing system, the bandwidth limits are 1000 Kbps, 300 Kbps and 100 Kbps respectively with a total at 1400 Kbps that is the same as the other two systems for the fairness of comparison. The configuration of Traffic Generator in Table 4 would meet the expected traffic throughput in Table 3. To test each system, there are three stages in our experiments. The whole process takes a total of 240 seconds. The payload size of each request sent by Traffic Generator will be based on the settings defined in Table 4. For the single slice scalable system, we follow the workflow for scaling as illustrated in Figure 6, and set the scale-out threshold to 50% and the scale-in threshold to 10% as in [29] [30]. • In the first stage, we will follow the configuration in Table 4 to send data to each system for 30 seconds. The requests of each application will be sent with different frequency and payload size. • In the second stage, we triple the number of user threads as shown in Table 4 and send data for 120 seconds. We simulate higher traffic in this stage. For the scalable system, the scale-out action will be triggered. • In the final stage, we return to the same configuration as the first stage for 90 seconds. During this stage, the scalable system will trigger the scale-in action back to its original status. Table 5 shows the throughput of each application type in each stage. These three systems had a similar result of the throughput and reach the values of expected traffic throughput shown in Table 3. Figure 7 shows the average response times of all applications in each system and Figure 8 shows the total CPU utilizations of three systems. Integrating the information from these two charts, we conclude that the multiple slicing system achieves the best response time at all stages. Also, its total CPU utilization is only slightly higher than the single slice system so it is an acceptable tradeoff. Overall, the performance of the multiple slicing system is better than those of the single slice systems whether it is equipped with scalability or not. Comparing the single slice system with the single slice scalable system, it is clear that the average response time of the single slicing system with scalability is better than the one without scalability. In the first stage, since the single slice scalable system must go through Load Balancer which is an additional VNF, the response time is longer than the single slice system. However, when the traffic load increases in the second stage, the response time of the single slice scalable system is similar to that of the single slice system. Moreover, the result of the single slice scalable system is even better in the final stage. This is because the single slice scalable system can deal with increasing traffic loads better than the single slice system. However, the total CPU utilization of the single slice scalable system is always higher than the other two systems due to the overhead of Load Balancer and scalability. The average response times of each application type in the single slice system over all testing stages are shown in Figure 9. The results show that the response time of each application type under the same system architecture only differs a little bit. Similar results are also exhibited in the other two systems. Figure 10 shows that the multiple slicing system gets the lowest response time regardless of the type of applications. We can thus conclude that the system architectures instead of the application types have deeper impact on the system performance. According to the above results, we speculate that implementing the horizontal scalability across the multiple slicing system may improve its performance and stability, which will be our future work. The research in [7] proved that network slicing can improve IoT/M2M scalability and fulfill different QoS requirements, which is also proved by our experimental results. However, the network slicing in [8] is based on SDN, while ours is based on NFV that has been adopted by the upcoming 3GPP 5G architecture. System Design, Implementation and Evaluation in Kubernetes In this section, we report our research results of building the NFV MANO framework with Tacker as NFVO/VNFM and Kubernetes [31] as VIM. In this design, OM2M IN instances are deployed as the containerized VNFs. Kubernetes is an open-source system for automating application deployment, scaling, and management. It provides a platform for deploying, managing, and scaling containerized applications across clusters of hosts. It works with a variety of container tools, including Docker. We construct only two slicing systems in this experiment including: 1) a single slice system and 2) a multiple slicing system. At the end of this section, we will compare and evaluate these two systems in terms of their average response time and CPU utilization. Figure 11 shows the general architecture of the two slicing systems. Each functional block of our systems has been presented before. Also, Tacker which is utilized as NFVO and VNFM has been introduced in Section 3. The only difference is that we now use Kubernetes instead of OpenStack as VIM. System Architecture Traffic Generator which is our design will simulate the same three types of traffic as our experiments in OpenStack. As depicted in Figure 12(a), the Single Slice System has only one network slice with one IoT platform as a containerized VNF. The only one network slice will deal with all types of IoT traffic. Figure 12(b) shows the architecture of the multiple slicing system where each network slice would handle a specific type of IoT services. Test Environment Setup Our test environment consists of two servers. Tacker and Kubernetes are each running on a server configured as shown in Table 6. Table 7 shows the virtual resource allocation of each containerized VNF in our environment. Experimental Results In this experiment, we use Traffic Generator to simulate three types of traffic and send the HTTP requests to each containerized VNF on the network slice. For the single slice system, we send all three types of traffic to the OM2M IoT Advances in Internet of Things platform directly. For the multiple slicing system, each type of IoT traffic will be sent to the containerized VNF on the corresponding network slice. The configuration of Traffic Generator is the same as the one used for Open-Stack as shown in Table 4 of Section 4. It will generate the expected traffic throughput required for each application type in this experiment as shown in Table 8. To test each system, there are three stages in our experiment. The whole process takes a total of 90 seconds. The payload size of each request sent by Traffic Generator will be based on the settings defined in Table 4 of Section 4. • In the first stage, we will follow the configuration in Table 4 to send data to each system for 30 seconds. The requests of each application will be sent with different frequency and payload size. • In the second stage, we triple the number of user threads as shown in Table 4 and send data for 30 seconds. We simulate higher traffic in this stage. • In the final stage, we return to the same configuration as the one in the first stage for 30 seconds. During this stage, the systems will approach stability. Because Kubernetes has its own scaling functions and policy for scalability, we only construct a single slice system and a multiple slicing system. Also, the time spent in the experiment for Kubernetes as VIM is different from the previous one for OpenStack as VIM. Since there was no need to do scalability, we shortened the total time of the experiment. Figure 13 shows the average response times of all applications in each system. It shows that the multiple slicing system achieves better response time at all stages although the response times of the two systems are similar in the first stage and the third stage. When the traffic load increases in the second stage, the response time of the multiple slicing system is half of that of the single slice system. On the other hand, as depicted in Figure 14 the CPU utilization of the multiple slicing system is always higher than that of the single slice system in all three stages because it has three network slices to handle different services. Note that the CPU utilizations of both systems are at their peak in the second stage due to the highest traffic load. On the other hand, because the first stage is the warm-up stage, the CPU utilizations of both systems are higher than those in the final stage when the systems become stable. Integrating the information from these two charts, we conclude that the performance of the multiple slicing system is better in general as its total CPU utilization is only slightly higher than that of the single slice system but it can achieve faster response time than the single slice system. Conclusions and Future Work In this paper, we propose three different slicing systems enabled by NFV, based on the MANO framework including: 1) a single slice system, 2) a multiple customized slices system and 3) a single but scalable network slice system to support IoT services. We utilize several open sources such as OpenStack, Tacker, Kubernetes, OM2M and RabbitMQ for constructing our system. In our system, To support different kinds of IoT services, we customize each network slice with a specific QoS. Moreover, we design a Master Node to monitor the CPU usage of each VNF and scale out or scale in VNFs on the slice according to this information. Also, Load Balancer is designed for the single slice scalable system to dispatch traffic fairly. In our experiment, we design Traffic Generator to simulate three types of IoT traffic including video, adaptive lighting and smart parking. The test traffic consists of three stages with different traffic loads. We measure the average response time and the CPU utilization of these three systems to identify the best system design. Comparing the results of these three systems, the multiple slicing system has the best performance among them. In addition, the single slicing system with scalability is more stable than the system without scalability with the tradeoff of higher CPU utilization. Combining the results of the two experiments, the multiple slicing system is the best system design. Although in our experiment with Kubernetes as the VIM, we only constructed the first two systems. The results also show that the performance of the multiple slicing system is better than that of the single slice system. In the future, we plan to construct a network slicing system with vertical scalability by adapting to changing QoS requirements dynamically. We also plan to experiment the horizontal scalability across multiple slices than just on a single slice. Moreover, constructing a hybrid system of horizontal and vertical scalability to meet more diverse requirements of IoT services is also a potential future research direction [32].	2020-10-28T18:31:11.752Z	2020-07-30T00:00:00.000	{ "year": 2020, "sha1": "775656aa5c67d238ec6d9e25e7701dd32ae5776f", "oa_license": "CCBY", "oa_url": "http://www.scirp.org/journal/PaperDownload.aspx?paperID=103057", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "72f8ff48623ea9b10b4f684846739a374b6f52d1", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Computer Science" ] }
228064228	pes2o/s2orc	v3-fos-license	Suppressing meta-holographic artifacts by laser coherence tuning A metasurface hologram combines fine spatial resolution and large viewing angles with a planar form factor and compact size. However, it suffers coherent artifacts originating from electromagnetic cross-talk between closely packed meta-atoms and fabrication defects of nanoscale features. Here, we introduce an efficient method to suppress all artifacts by fine-tuning the spatial coherence of illumination. Our method is implemented with a degenerate cavity laser, which allows a precise and continuous tuning of the spatial coherence over a wide range, with little variation in the emission spectrum and total power. We find the optimal degree of spatial coherence to suppress the coherent artifacts of a meta-hologram while maintaining the image sharpness. This work paves the way to compact and dynamical holographic displays free of coherent defects. Introduction Artificial metasurfaces, comprised of a two-dimensional (2D) array of subwavelength scatterers, have shown unprecedented ability in controlling optical wavefront and converting conventional bulky optical elements into planar thin films 1-3 . One prominent example is the metasurface hologram (meta-hologram) [4][5][6][7] . An ultrathin metasurface is capable of reconstructing a three-dimensional (3D) holographic image with a high spatial resolution and large viewing angles, while suppressing high-order diffraction [8][9][10][11][12] . Very recently, multi-color, multiplexed, and dynamic metaholograms have been proposed and demonstrated, illustrating a great potential in information processing, 3D display, high-density data storage, and optical image encoding [13][14][15][16][17][18][19][20] . Despite of these remarkable advances, the road to practical applications of meta-holograms is hindered by coherent artifacts. Such artifacts originate from near-field interactions of subwavelength scatterers (meta-atoms), fabrication defects and phase dislocations, causing image distortion and degradation 6,21 . While coherent artifacts and speckle noise are well-known issues for conventional holography, they are more significant in regard to meta-holography, as close packing of meta-atoms enhances the cross-talk and fabrication of nanoscale features is susceptible to error. Such artifacts cause severe distortions of holographic images, which are extremely difficult to correct. Recently, machine-learning-based optimization techniques have been applied to high-performance metasurface design [22][23][24] . They require a large amount of training data, which are difficult and expensive to acquire for large-scale meta-holograms. Moreover, the coherent artifacts caused by random fabrication imperfections cannot be removed by optimizing the meta-hologram design. An alternative way of suppressing coherent artifacts is adjusting the spatial and/or temporal coherence of illuminating light 25 . In general, it is more efficient to suppress speckle noise by reducing spatial coherence than temporal coherence 26 . While lowering the temporal coherence with broadband illumination provides spectral compounding, the different wavelengths generate different radial scalings of speckle patterns resulting in a deficient suppression of coherent artifacts 27 . Previously, lowering the spatial coherence of illumination has been widely used for the suppression of speckle noise in conventional holography 28 . It is done by either increasing the spatial coherence of a light-emitting-diode (LED) with spatial filtering [29][30][31] , or decreasing the spatial coherence of a laser with moving elements and time integration [32][33][34][35] . While the former technique suffers from severe power loss, the latter requires long exposure time. Since reducing the spatial coherence would blur the image and reduce the depth of field, a precise tuning of the spatial coherence is required. To eliminate coherent artifacts of meta-holograms, we resort to a degenerate cavity laser (DCL) with tunable spatial coherence for illumination. The DCL provides a wide tuning range of spatial coherence with little power loss 36,37 . Its fast decoherence enables a short exposure time for high-speed imaging 38 . Furthermore, the emission spectrum of the DCL does not change during the spatial coherence tuning, avoiding the spectral dispersion of the meta-hologram. By fine-tuning the spatial coherence of the DCL, we find the optimal degree that suppresses all sorts of artifacts without a significant blurring of the holographic image. Our scheme works efficiently for different types of meta-holograms, providing a general method for artifact-free holographic display. Coherent artifacts created by meta-holograms We design and fabricate a metasurface hologram as shown in Fig. 1a. It is made of Silicon nanopillars on top of a glass substrate (see "Methods" for the design and fabrication processes). The phase modulation of the meta-hologram is achieved via resonant scattering of individual nanopillars (meta-atoms). By tuning the pillar diameter D, the scattering resonance frequency is varied and the phase response φ at the illumination wavelength is changed. In the metahologram design, φ(D) is calculated for a single pillar with periodic boundary conditions. To create a holographic image in the far-field, the near-field phase pattern φ H is computed with an iterative phase retrieval (IPR) algorithm (see "Methods" for details). Then, the inverse mapping function D (φ) determines the spatial variation of the nanopillar diameter D to obtain the designed phase distribution φ H . Since the nanopillars are densely packed, the near-field interactions among neighboring nanopillars are significant. In the hologram design, the periodic boundary conditions used in the phase calculation φ(D) correspond to the assumption that all neighboring pillars have an identical diameter D. This is not true in the actual metahologram, where D varies spatially. The near-field interactions between nanopillars with different diameters differ from the ones with the same diameter. Such difference causes the actual phase response (φ A ) to deviate from the designed one (φ H ). To reduce this deviation, we adopt a unit cell with 2 × 2 identical nanopillars, so that some of the neighboring pillars have the same diameter and their couplings better agree with the calculation with periodic boundary condition. However, the increase of the unit cell size reduces the maximal viewing angle to 99°. To keep a relatively large viewing angle, we refrain from further enlarging the unit cell. Albeit weaker than the case of single-pillar unit cell, the cross-talk effects are still strong for the 4-pillar unit cell, as confirmed by a numerical simulation of a small meta-hologram with 8 × 8 unit cells. The actual phase modulation φ A in Fig. 1b is notably different from the designed pattern φ H . Such difference leads to a severe distortion of the holographic image, as observed experimentally in Fig. 1c. The seemingly-random intensity fluctuation is reproduced with another fabricated meta-hologram of identical design in Fig. 1d. Magnified images are available in the Supplementary Materials. Therefore, the artifacts are primarily due to deterministic crosstalk among the meta-atoms. The resulting contrast of intensity fluctuations in the meta-holographic image is ∼0.3, much higher than the typical speckle contrast of classical holographic images. Such cross-talk is very difficult to amend, because the meta-hologram in Fig. 1a is comprised of 128 × 128 = 16,384 unit cells, i.e., in total 16,384 × 4 = 65,536 nanopillars (meta-atoms). To accurately account for the interactions among all nanopillars of varying size, the phase response of the entire metasurface has to be calculated, a task which is computationally demanding. Any iterative optimization of the hologram configuration requires simulating the entire metasurface repetitively, which is beyond standard computing capabilities. It is neither practical to apply machine learning to this case, as an extensive simulation of such large meta-holograms with varying parameters is needed to train an artificial neural network. In addition to the cross-talk of meta-atoms due to their near-field interactions, there are two different sources for meta-hologram artifacts. Due to the subwavelength size of the Silicon pillars, structural defects are introduced unintentionally during the fabrication of meta-hologram, as seen in the inset of Fig. 1a. Such defects induce unwanted light scattering and interference, producing additional artifacts in the holographic image. Furthermore, optical vortices are generated in the holographic image due to the creation of phase dislocations in the design of digital hologram. Such defects can be eliminated by incorporating complementary algorithms in the IPR [39][40][41][42][43] (see "Methods"). Alternatively, they are eliminated by full-field (amplitude and phase) modulation with a meta-hologram 44,45 . Degenerate cavity laser with tunable spatial coherence To suppress coherent artifacts, we adjust the spatial coherence of illumination with a degenerate cavity laser (DCL). The DCL has a self-imaging configuration 46 , as shown in Fig. 2a. Since many transverse modes have a nearly degenerate quality factor, they can lase simultaneously and independently to reduce the spatial coherence of the emission. By tuning the cavity away from the degenerate condition (see "Methods"), the number of transverse lasing modes decreases, and the degree of spatial coherence increases. Figure 2b shows the near-field (top row) and far-field (bottom row) intensity patterns of the laser emission. The near-field patterns at the DCL output coupler consist of bright spots, each corresponding to a transverse lasing mode. As the cavity approaches the degenerate condition, the number of spots (modes) increases. The diffracted beams from neighboring spots do not interfere, indicating that the modes are mutually incoherent. The number of independent lasing modes N is estimated from the intensity contrast of a speckle pattern generated by a static diffuser placed outside the laser cavity (see "Methods" and Supplementary Materials). As N decreases from ∼300 to ∼1, the emission power is merely reduced by 40% from 108 mW to 64 mW (see Supplementary Materials). Furthermore, the emission spectrum remains approximately the same with a full-width-at-half-maximum (FWHM) of about 3 nm, indicating that the temporal degree of coherence does not change (see Supplementary Materials). This effectively avoids the influence of chromatic aberration. In contrast to the spotted near-field pattern, the laser emission exhibits a smooth profile at the far-field. It is composed of an incoherent superposition of Gaussian beams propagating in slightly different directions from individual spots at the near-field. The smooth intensity Actual phase response φ A from a numerical simulation of the entire metasurface, showing a significant deviation from the designed one due to near-field interactions among neighboring nanopillars. c, d Holographic images of a star object generated by two fabricated meta-holograms with the same design shown in (a). Their intensity fluctuations are nearly identical, indicating that the fluctuations result mainly from deterministic interactions among meta-atoms. Optical vortices are already eliminated from the computer-generated hologram. The illumination source is a monochromatic laser at wavelength λ = 1064 nm, which has a high spatial and temporal coherence distribution ensures a homogeneous illumination of the meta-hologram which is placed at the far-field of the DCL. The holographic image is created in the far-field of the meta-hologram. In the case of coherent illumination, the emission from a single transverse lasing mode illuminates the meta-hologram, as sketched in Fig. 2c. With partially coherent illumination in Fig. 2d, mutually incoherent lasing modes illuminate the meta-hologram with different angles of incidence and generate holographic images that are laterally shifted at the far-field. The number of distinct images is given by the effective number of independent spatial modes N E in illumination, which is equal to the ratio of the area of the hologram to the coherence area of the illuminating light (see "Methods"). An intensity sum of N E images will average out the intensity fluctuations due to coherent artifacts. However, the averaging also blurs the image and impairs the spatial resolution. Hence, the degree of spatial coherence must be optimized to suppress coherent artifacts without significantly degrading the image resolution. Suppression of holographic artifacts To demonstrate the capability of our method in suppressing all sorts of coherent artifacts, we design one set of meta-holograms with the standard Gerchberg-Saxton IPR algorithm 47 . The holographic images of this set contain many dark spots due to phase dislocations (optical vortices). The top row of Fig. 3 shows the holographic images, taken with varying degrees of spatial coherence of the DCL illumination. N E is the effective number of spatial modes that illuminate the meta-hologram and generate laterally shifted holographic images. When the spatial coherence is high (N E ≅ 1), the holographic image is full of coherent artifacts generated by near-field meta-atom interactions, fabrication defects, and phase dislocations. Lowering the spatial coherence by increasing N E to 21 suppresses the intensity fluctuations, resulting in a nearly uniform holographic image. A further increase of N E to 30, however, results in a notable reduction of the edge sharpness, as seen in the 1D intensity profile across an edge of the star image in Fig. 3. We also test our method with meta-holograms free of phase dislocations. Despite the absence of optical vortices in the holographic images, intensity fluctuations are still significant, as seen in Fig. 1c Fig. 2 Tuning the spatial coherence with a degenerate cavity laser (DCL). a Schematic of the DCL comprising of a vertical external cavity surface emitting laser (VECSEL) module (VL), two lenses (L1, L2), and an output coupler (OC). The far-field DCL emission is projected by imaging optics (L3) onto the meta-hologram (MH), which creates a holographic image (HI) at its far-field. L1 is gradually translated along the cavity axis to break the cavity degeneracy condition, so that the number of transverse lasing modes N decreases and the spatial coherence of the total emission increases. b Near-field (top row) and far-field (bottom row) intensity patterns of the total emission with a varying number N of transverse lasing modes. With increasing N, there are more bright spots in the near-field (each corresponding to an independent lasing mode), and the far-field pattern becomes larger. c Schematic of a single lasing mode illuminating the meta-hologram, creating intensity fluctuations due to coherent artifacts. d Schematic of two mutually incoherent lasing modes illuminating the meta-hologram with different angles, creating laterally shifted and mutually incoherent holographic images. An incoherent (intensity) sum of the two images reduces the intensity fluctuations In addition to the resonant phase meta-holograms, our method is applicable to geometric Pancharatnam-Berry (PB) phase meta-holograms 1,48,49 . These holograms are also cleared of phase dislocations originating from the phase encoding process. A scanning electron microscope (SEM) image of the fabricated hologram is presented in the Supplementary Materials. The bottom row of Fig. 3 shows the holographic images recorded with the DCL illumination. Again, we observe intensity fluctuations under high spatial coherence illumination. In the absence of optical vortices, the artifacts result mainly from nearfield interactions of meta-atoms. The intensity fluctuations are not as strong as those with optical vortices, thus a small decrease of the spatial coherence is sufficient to make the image smooth. The edges get blurred with a further reduction of the spatial coherence. The results of both types of meta-holograms confirm that the finetuning of the DCL spatial coherence is critical in achieving an optimal illumination condition where the holographic image is nearly free of coherent artifacts and remains relatively sharp. Optimal degree of spatial coherence To quantitatively assess the holographic image quality, we evaluate the signal-to-noise ratio (SNR) and edge sharpness. Experimentally, we collect the data of five meta-holograms free of phase dislocations. The metaholograms generate the same holographic image of a star, as shown in the inset of Fig. 4a. The SNR is defined as I h i/σ, where I h i is the average intensity within the central square marked in the inset of Fig. 4a, and σ is the standard deviation of the intensity fluctuation in this region. We average the SNR over five meta-holographic images, and plot its value versus the effective number of spatial modes N E . As seen in Fig. 4a, the SNR increases monotonically with N E . In logarithmic scales, the data points follow a straight line of slope 1/2, indicating that the SNR scales as ffiffiffiffiffiffi N E p . The edge sharpness is estimated from several 1D intensity profiles of the holographic image across different edges, as marked by green dotted-dashed lines in the inset of Fig. 4a. The sharpness is evaluated by estimating the slope of the edge, between two intensity points corresponding to 10% and 90% of the maximum intensity, S = (r 10% − r 90% ) −1 (see the inset of Fig. 4b). Figure 4b shows the edge sharpness S averaged over multiple edges of five holographic images. As the effective number of spatial modes N E grows, S drops monotonically. Next we model the dependence of S on N E . For N E = 1, the edge intensity profile is given by convolution of the ideal step function The coherent artifacts seen with high spatial coherence illumination (first column) gradually disappear, as the spatial coherence is lowered by increasing the effective number of spatial modes N E that illuminate the metahologram (second column). Further decrease of the spatial coherence (increase of N E ) notably blurs the image and reduces the edge sharpness (third column). The fourth column shows the 1D intensity profile through a cut of the holographic images marked by white dotted line in the first three columns with the point spread function (PSF) of the holographic imaging setup. The width w P of the PSF determines the spatial resolution, and is inversely proportional to the lateral dimension of the meta-hologram. For N E > 1, there are N E laterally shifted holographic images created by the meta-hologram, and an incoherent summation of all images broadens the edge intensity profile. The broadening depends on the width of the DCL near-field emission pattern (Fig. 2b), and is proportional to ffiffiffiffiffiffiffiffiffiffiffiffiffiffi N E À 1 p . As a result of convolution, the sharpness is given by where C 0 is a prefactor, C 1 is a scaling constant relating the lateral shift of a holographic image to the tilt of incident angle of an illuminating beam (see Supplementary Materials for a complete derivation). Once the values of C 0 and C 1 are determined by fitting the experimental data with the expression of S(N E ), the theoretical prediction of the edge sharpness (green curve in Fig. 4b) captures the measured dependence of S on N E . Finally, we identify the optimal degree of spatial coherence for illuminating a meta-hologram and find its dependence on the image resolution. To this end, we fabricate another set of meta-holograms that produce holographic images of a USAF resolution test chart. Figure 5a shows three holographic images with different degrees of spatial coherence illumination. The image quality is assessed by the contrast to noise ratio (CNR), which is defined as where I S h i and σ S denote the average intensity and the standard deviation of the intensity fluctuation in the bright region, respectively. I B h i is the average intensity of the dark background. The numerator of the CNR, 1 − I B h i= I S h i, gives the intensity contrast between bright and dark regions, while the denominator σ S = I S h i characterizes the intensity fluctuation (noise) in the bright region. Overall, the CNR describes the resolvability of a feature of interest (bright) in a given background (dark). Figure 5b shows the measured CNR varying nonmonotonically with the effective number of spatial modes N E for three different feature sizes. As N E increases, the CNR first grows and then drops. It reaches a maximal value at an intermediate N E , indicating that there is an optimal degree of spatial coherence for illumination. When the feature size is large, the CNR reaches the maximum at a relatively large N E . Since the intensity contrast (in the numerator of CNR) remains high for a relatively wide range of N E , the CNR is determined mainly by the intensity fluctuation (in the denominator), which is smaller at larger N E . As the feature size decreases, the maximum CNR shifts to a lower N E . That is because, in resolving small features, the intensity contrast becomes more significant and is higher at smaller N E due to less blurring. However, the maximal value of the CNR is less than that for a large feature size, because the intensity fluctuations are stronger. Therefore, the optimal degree of spatial coherence increases with the image resolution. Figure 5c shows the optimal number of independent spatial modes N ðmaxÞ E (number of laterally shifted holographic images) required to reach the maximum CNR as a function of the spatial frequency (inverse of spatial resolution) in the USAF test chart. As the spatial frequency increases, the feature size decreases, and N Discussion In this work, we tune the spatial coherence of a degenerate cavity laser (DCL) to suppress strong coherent artifacts created by metasurface holograms. Compared to the conventional method of lowering the spatial coherence of a laser by a moving diffuser, our approach has several distinct advantages: (i) The precise, continuous tuning of the DCL spatial coherence allows to reach the maximal contrast to noise ratio (CNR) for any desired spatial resolution. (ii) The tuning is energy efficient, and does not introduce a significant power variation. (iii) The spectral width of the DCL emission (degree of temporal coherence) remains constant during the spatial coherence tuning, which is important for metaholograms with strong dispersion. (iv) No pre-or postprocessing procedures are needed in our method for coherent artifacts suppression. (v) Fast lasing dynamics leads to rapid decoherence of the DCL emission, thus enabling high-speed meta-holography. In comparison to other incoherent light sources, the spectral radiance (photon degeneracy number) of our DCL exceeds that of a superluminescent diode (SLD) by one order of magnitude and a light-emitting diode (LED) by six orders of magnitude (see Supplementary Materials). Such high brightness is critical to dynamic imaging with short integration time. Tuning the spatial coherence is more appropriate to suppress coherent artifacts than tuning the temporal coherence, as illustrated in Fig. 6. When illuminated by broadband light with low temporal coherence but high spatial coherence, a meta-hologram creates multiple images whose size scales with the wavelength (see Fig. 6b). An incoherent sum of these images produces a radially extended image with strong edge blurring, as confirmed numerically in Fig. 6e Optimal degree of spatial coherence. a Three holographic images of USAF test charts obtained with the same meta-hologram and illuminated by the DCL with varying degrees of spatial coherence. As the effective numbers of spatial modes N E increases, the spatial coherence of the illumination decreases and the coherent artifacts are suppressed, however, the image resolution is impaired. b Contrast to noise ratio (CNR) versus the effective number of modes N E for three spatial frequencies in the test charts. The CNR first increases with N E , reaches a maximum (circled in red) and then decreases. The maximum of the CNR shifts to a lower N E for higher spatial frequencies (smaller feature size). c The optimal number of spatial modes N ðmaxÞ E at the maximal CNR decreases, as the spatial frequency (inverse of spatial resolution) increases. The purple curve is the theoretical prediction of our model image dilation for different wavelengths occurs along a diagonal direction, causing an inhomogeneous edge blurring (see Fig. 6d). In contrast, the edge blurring is less severe and more homogeneous when lowering the spatial coherence of illuminating light for both on-axis and offaxis holography (see Fig. 6a, c). In addition, the temporal coherence tuning is usually done by spectral filtering of a broadband source, which causes a notable change of illumination power. While the coherent artifacts are strongly suppressed by our method, they cannot be completely removed by lowering the spatial coherence, otherwise, the image blurring would be too severe. Full removal of all coherent artifacts with little loss of spatial resolution is challenging, but may be possible in the future via a combination of minimizing cross-talk and phase dislocation by optimizing the metasurface structure with machine-learning and inverse design, reducing the fabrication defects with highprecision nanofabrication, and removing the residual artifacts with a slight decrease of the spatial coherence of illuminating light. In summary, our scheme can rapidly and efficiently suppress all coherent artifacts created by different types of meta-holograms. It paves the way for the applications of meta-holograms in dynamic display, augmented reality, optical storage, beam multiplexing, nonlinear holography, and optical manipulation. Digital meta-hologram design We design two types of metasurface holograms. In the first one, the phase modulation is obtained via resonant scattering of silicon nanopillars with varying diameter. The second type is based on a geometric Pancharatnam-Berry (PB) phase modulation induced by silicon nanofins of different in-plane orientation (meta-atoms). Using a commercially available finite element method (FEM) solver (COMSOL Multiphysics), we calculate the phase response φ of a single nanopillar with diameter D and a nanofin with orientation angle θ. Periodic boundary conditions are applied, thus neighboring nanopillars (nanofins) are assumed to have an identical diameter D (orientation angle θ). By varying (θ), we obtain the mapping φ(D) [φ(θ)]. We use an iterative phase retrieval (IPR) algorithm to find the near-field phase modulation creating a far-field holographic image. The standard method based on the Gerchberg-Saxton (GS) algorithm 47 generates optical vortices (phase dislocations) in the holographic image. To eliminate such artifacts, the GS algorithm is modified with an initial spherical phase front 50 Fig. 6 Comparison of spatial and temporal coherence tuning on holographic image quality. a-d Schematic of holographic images created by illuminating light with low spatial coherence and high temporal coherence (a, c), or with low temporal coherence but high spatial coherence (b, d) for on-axis holography (a, b) and off-axis holography (c, d). The optical axis is marked by a red dot. Lowering the temporal coherence results in stronger and non-uniform edge blurring than lowering the spatial coherence. e Numerically calculated meta-holographic images created by plane wave illumination at wavelengths λ = 965 nm, 1055 nm, 1455 nm. Pseudo-colors are used to label the wavelengths. The image size increases with wavelength, and an intensity sum of these images produces a radially extended image followed by a simulated annealing (SA) routine 39,41 . Finally, we encode the near-field phase profile φ H , unit cell by unit cell, in a metasurface comprising of 128 × 128 unit cells. Each unit cell contains 2 × 2 nanopillars (nanofins) of the same size (orientation). The nanopillar (nanofin) diameter D (orientation angle θ) in every unit cell is chosen from the inverse mapping D(φ) [θ(φ)]. For the nanopillar hologram, the 2π phase modulation is achieved by varying D from 142 nm to 366 nm (see Supplementary Materials). In the nanofin hologram, all nanofins have an identical size but varying orientation. Each fin has a length of 393 nm, a width of 82 nm, and a thickness of 600 nm. The geometric phase φ is dictated by the in-plane orientation angle θ: φ = ±2θ, where the ± signs correspond to left-and right-circular polarizations of incident light. With θ varying from 0 to π, the phase φ modulation covers a 2π range. When illuminated by the linear polarized emission from the DCL, the nanofin hologram generates two images of left and right circular polarizations at different far-field locations. While the design of the meta-hologram phase profile φ H is done by tiling unit cells with the pre-calculated phase response of individual meta-atoms, the actual phase response φ A of a small meta-hologram with 8 × 8 unit cells is calculated by the FEM with absorbing boundary conditions, and shown in the right panel of Fig. 1b. Meta-surface fabrication The silicon (Si) metasurfaces are fabricated with electron-beam lithography and reactive ion etching. A 600 nm thick amorphous Si film is deposited on a glass substrate using an electron beam evaporator (SKE_A_75). Then, a 100 nm thick electron-beam resist (MicroChem PMMA [polymethyl methacrylate] A2) is spin-coated onto the Si film and patterned with the electron beam writer (Raith E-line, 30 kV). After development in a MIBK & IPA (1:3) solution, 15 nm thick Chromium (Cr) is deposited on the sample using electron beam evaporation (SKE_A_75) and the inverse nano-pattern is transferred to the Cr layer by a lift-off process in a remover PG (Micro Chem). By etching the Si with Sulfur hexafluoride (SF6) and Fluoroform (CHF3) in a flow of 5 sccm and 50 sccm, respectively, the nano-patterns are transferred to the Si membrane. Finally, the Cr mask is removed by immersing the samples in a Cr etchant solution (Aldrich Chemistry) for 30 minutes. Spatial coherence tuning of the DCL The coherence level of the DCL is tuned by translating a lens (L1 in Fig. 2a) inside the cavity along the longitudinal axis. The lens is mounted on a mechanical translation stage with micrometer resolution (Thorlabs MBT616D). When the lens L1 is accurately positioned (Δz = 0 μm) to satisfy the cavity self-imaging condition, lasing occurs in many transverse modes with nearly degenerate loss, and the total emission has a low spatial coherence. Moving the lens L1 from Δz = 0 μm breaks the degeneracy condition and reduces the number of transverse lasing modes. At Δz = 300 μm, nearly all transverse modes stop lasing except one and the spatial coherence of emission is high. By gradually changing Δz from 0 μm to 300 μm, we can continuously vary the number of transverse lasing modes, and accurately tune the degree of spatial coherence from low to high. Characterization of spatial coherence To measure the number of independent transverse lasing modes N in the DCL, we direct the lasing emission to a spatial coherence measurement setup. It consists of two lens (L3, L4) with identical focal length f, which are arranged in a 4f configuration. A ground glass diffuser (Thorlabs DG10-600) is placed in between L3 and L4 at the mutual focal plane (see Supplementary Materials for a schematic and more details). The speckle pattern generated by the diffuser is measured by a CCD camera at the back focal plane of the second lens L4. The intensity contrast C of the speckle pattern is defined as C = σ I = I h i, where I h i is the mean intensity and σ I is the standard deviation of intensity fluctuation. C is related to the number of independent transverse lasing modes N by . For different detunings (Δz) of the DCL, the number of transverse lasing modes is estimated from the measured speckle contrast: N ¼ 1=C 2 . In order to validate the mutual incoherence of the bright isolated spots in the near-field emission pattern of DCL (Fig. 2b), we measure the spatial pattern of the output beam as it propagates away from the DCL. The diffraction of emission from individual spots causes a spatial overlap of neighboring ones, but no interference fringes are observed in the time-integrated emission pattern, indicating these lasing spots are mutually incoherent. In the holographic imaging setup, the diameter L I of the illuminating beam is larger than the lateral dimension L H of a meta-hologram to ensure a uniform intensity illumination. The angular width of the illuminating beam ΔΘ I is inversely proportional to the spatial coherence length L C : ΔΘ I / 1=L C . The ratio between the illuminating beam area A I / L 2 I and the coherence area A C / L 2 C gives the number of independent spatial modes: Since the illuminating beam area is larger than the meta-hologram area A H ¼ L 2 H , the effective number of modes N E within the meta-hologram is smaller than N. The effective number of modes N E , interacting with the meta-hologram and producing independent holographic images, is given by the ratio of the meta-hologram area and the illumination coherence area: where ΔΘ H / 1=L H is the diffraction angle of the metahologram due to its finite size. The effective number of spatial modes in illumination N E gives the number of independent holographic images generated at the farfield. In our setup, the ratio N=N E ¼ A I =A H is found to be approximately 8.	2020-12-10T02:15:51.994Z	2020-12-08T00:00:00.000	{ "year": 2020, "sha1": "b51d9b1b5e65e99d3776f5a1cb19cdbbb5b240ee", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41377-021-00547-0.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "ef5ed9a2b5467cc9b957b3c8968d4e95dd58a63f", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics", "Medicine" ] }
253566208	pes2o/s2orc	v3-fos-license	Nomenclatural Synopsis, Revised Distribution and Conservation Status of Ranunculus gracilis (Ranunculaceae) in Italy Ranuculus gracilis is endemic to the SE Euro-Mediterranean area and its presence in Italy is controversial. Based on analysis of the relevant literature, field surveys and examination of herbarium specimens, a revised distribution of this species in Italy is presented and its conservation status is assessed. Ranunculus agerii, described by Antonio Bertoloni from Bologna (Emilia-Romagna, Northern Italy), and R. schowii, described by Vincenzo Tineo from Vittoria and Terranova (Sicily), usually regarded as synonyms of R. gracilis, are here lectotypified and their taxonomic status discussed. Thanks to our study, the presence of R. gracilis in Italy is confirmed and, now, it is reported in a national conservation framework. Introduction Ranunculus L. is the largest genus in the family Ranunculaceae Juss., with a cosmopolitan distribution, consisting of about 1200 species (including also ca. 600 agamospecies) [1,2]. Based on morphological and molecular data, the genus Ranunculus was divided into two subgenera (subg. Auricomus and subg. Ranunculus) and 17 sections [2]. In Italy, the genus Ranunculus comprises 112 taxa (species, subspecies and agamospecies), of which 33 are endemic (four are extinct) and one is an alien species [3][4][5]. Ranuculus gracilis was described in 1814 by Edward Daniel Clarke from the East Aegean island of Kos (Greece) [6] and belongs to R. subg. Ranunculus sect. Ranunculastrum DC. [2]. This species is endemic to the SE Euro-Mediterranean area and it is distributed in Italy, the Balkan Peninsula, Turkey and Georgia [7,8]. Contrary to what is reported by Euro+Med Plantbase [7] and POWO [8], the presence of R. gracilis in Italy is controversial, i.e., [4,9]. According to the latest Italian Flora [9], it is present in a few localities in Sicily and Calabria, no longer recorded in Emilia-Romagna and cultivated in Umbria. On the contrary, following the Italian Checklist [4], it is recorded by mistake in Piemonte, Emilia-Romagna, Umbria, Sicily and doubtfully occurring in Puglia. Two species, described from Italy, are usually regarded as synonyms of R. gracilis (e.g., [9][10][11][12]): R. agerii Bertol. and R. schowii Tineo (with doubt). Ranunculus agerii was described in 1819 by Antonio Bertoloni from the areas around Bologna (Emilia-Romagna, northern Apennines, Italy) [13] and it is currently regarded as a synonym of R. gracilis (i.e., [7][8][9]12,[14][15][16][17]). Ranunculus schowii was described by Vincenzo Tineo in Gussone [18] from Vittoria and Terranova (Sicily) and regarded as a dubious synonym of R. gracilis or R. agerii by some authors (i.e., [10][11][12]) or as a synonym of R. monspeliacus L. subsp. monspeliacus (i.e., [8]). The purpose of this study is to critically review the presence of R. gracilis in Italy and to understand the taxonomic identity of R. agerii and R. schowii, two names that turned out to be, to the best of our knowledge, not yet typified. The present contribution is part of an ongoing project promoted by the Italian Botanical Society, aimed at recognizing and typifying all the taxa described from Italy, in order to increase their systematic knowledge and promote further studies [19][20][21][22]. Materials and Methods This study is based on an extensive analysis of relevant literature, field surveys and examination of herbarium specimens (including the original material) preserved in APP, BOLO, BR, CAT, CHE, CLF, DR, E, FI, G, GOET, JE, MW, NAP, P, PI, PAL, W and WU (the acronyms follow [23]). We performed a survey for original material for the name R. agerii at BOLO where Bertoloni's main collection is housed and at FI, G, NAP and PAL to trace the original material of the name R. schowii (see [24,25]). The original material for the name R. gracilis was searched at A, BM, CAN, CGE, ECON, GH, FI, K and SWN. The type designations herein follow the Shenzhen Code ( [26], hereafter ICN). The revised Italian distribution of R. gracilis is based on examination of herbarium specimens. The distribution data and the occurrence status are given for the Italian administrative regions according to Bartolucci et al. [4]. The Long and Controversial History of R. gracilis in Italy Over the years, the presence of R. gracilis in Italy has been controversial due to the confusion in its identification and to the unclear taxonomic relationships with some currently not accepted species [7,8], such as R. agerii Bertol., R. schowii Tin. and R. chaerophyllos L. The first herbarium samples collected in Italy that can be referred with certainty to R. gracilis date back to the 16th century. There are two specimens without collection locality preserved in the "En Tibi" herbarium, made by Francesco Petrollini in Bologna around 1558, kept in Leiden (an image of the specimen is available at https://data.biodiversitydata.nl/ naturalis/specimen/L.2110949 (accessed on 10 September 2022)) and in the controversial "Cibo" herbarium kept in Rome in Biblioteca Angelica, recently also attributed to Petrollini [27,28]. According to Stefanaki et al. [28], it is evident that many specimens in the "En Tibi" herbarium were collected in the area of Bologna, where Petrollini had his place of residence. Towards the end of 1500, this plant was also collected by Nicolas Ager (or Agerius) in the Bologna hills. He sent samples to Jean Bauhin, who gave a brief description of this plant in the Phytopinax published by Caspar Bauhin as "Ranunculus racemosa radice Io. Bauhini . . . Reperitur in montibus Bononiensibus" [29]. Later, it was reported by the Bauhin brothers as "Ranunculus grumosa radice folio ranunculi bulbosi . . . Hic Ranunculus agris Bononiensibus familiaribus est, & à D. Agerio collectus" in the Prodromos theatri botanici [30], "Ranunculus grumosa radice folio ranunculi bulbosi" in the Pinax theatri botanici [31], "Ranunculus racemosa radice . . . Agerio siccam dedit pro Ranunculo Chelidoniae radice" in the Historia plantarum universalis [32] and by Parkinson [33] as "Ranunculus grumosa radice Bononiensis". Later, Linnaeus [34], mistakenly included the polynomial published by C. Bauhin in the Prodromus [30] and in the Pinax [31] in R. chaerophyllos L. Antonio Bertoloni was the first to accurately describe [13] the plant from the areas around Bologna (Emilia-Romagna) as "Ranoncolo Bolognese", dedicating it to Nicolas Ager, with the name R. agerii. After the description, R. agerii was treated as synonym of R. chaerophyllos L. by Arcangeli [35] and Cesati et al. [36], a name of uncertain application [37]. Later, Fiori et al. [10] and Fiori [11] re-evaluated R. agerii as a good species, recording it not only for its locus classicus, but also for Sicily and quoting R. gracilis and, with doubt, R. schowii Tineo, only in [11] as synonyms. Pons [38] recorded R. agerii (syn. R. gracilis) for several localities around Bologna in Emilia-Romagna and for Catania in Sicily. Tutin [14] and Tutin and Akeroyd [17] in Flora Europaea quoted R. gracilis (syn. R. agerii) for Italy and Sicily. Zangheri [39] reported R. gracilis (syn. R. agerii) for Sicily and as naturalized in Northern and Central Italy. Pignatti [15] reported R. gracilis (syn. R. agerii) in Sicily and no longer recorded for Emilia-Romagna. Greuter et al. [40] in the Med-Checklist quoted R. gracilis (syn. R. agerii) for Sicily and as doubtfully native in Italy. Jalas and Suominen [16] quoted R. gracilis (syn. R. agerii) for Calabria and Sicily and with doubt in Emilia-Romagna. Peruzzi and Passalacqua [41] reported R. gracilis for the Balkan Peninsula, Turkey, and Crete, while, the Italian records of Calabria and Sicily, should be referred to R. monspeliacus L. subsp. aspromontanus (Huter, Porta & Rigo) Peruzzi & N.G.Passal. Conti et al. [42] reported R. gracilis without synonyms as doubtfully occurring in Italy in Piemonte, Emilia-Romagna and Sicily. In the same year, Scoppola and Spampinato [43] recorded R. gracilis (syn. R. agerii) for Sicily and as indicated by mistake in Emilia-Romagna. Later, Conti et al. [44] reported R. gracilis as indicated by mistake in Italy, updating the occurrence status in Conti et al. [42]. Recently, Pignatti et al. [9] quoted R. gracilis (syn. R. agerii) as present in a few localities in Sicily, Calabria, no longer recorded in Emilia-Romagna and cultivated in Umbria. On the contrary, according to Bartolucci et al. [4], in the updated checklist of Italian vascular Flora, the species was recorded by mistake in Piemonte, Emilia-Romagna, Umbria, Sicily and as doubtfully occurring in Puglia. Recently, Guarino and La Rosa [45] in the Digital Flora of Italy included in the 4 th volume of Flora of Italy [46], recorded R. gracilis for Calabria, alien in Umbria and as doubtfully occurring in Sicily and Emilia-Romagna. Figure 1B). Typification of the Names Nomenclatural notes: Antonio Bertoloni [13] described R. agerii, providing a detailed description, quoting a precise collection locality and citing an illustration "Tab VI". In BOLO, where Bertoloni's main collection is housed, we traced only one herbarium sample with two mounted individuals collected in 1818 on Monte Donato ( Figure 1) that can be considered as part of the original material, as well as the illustration "Tab VI" cited in the protologue (Figure 1) (Art. 9.4 of the ICN). The herbarium sample kept in BOLO is complete, well conserved and agrees with the protologue and is selected here as a lectotype for the name R. agerii. Taxonomic notes: based on the original material studied, R. agerii should be regarded as a heterotypic synonym of R. gracilis. Figure 2). Nomenclatural notes: Vincenzo Tineo in Gussone [18] described R. schowii, providing a detailed description, quoting a precise collection locality and citing an unpublished illustration "Tin. ined.". In order to trace the original material, we checked the PAL herbarium, where Tineo's main collection is housed. We also searched in FI, G, NAP and P, where duplicates by Tineo's collections are kept. Lojacono Pojero [48] wrote that he saw the only authentic specimen of R. schowii in H. Pan. (i.e., Herbarium Panormitanum), today PAL. In the Herbarium Mediterraneum Panormitanum (PAL), this sample is no longer present (G. Domina, pers. comm.). In NAP (ex-Herbarium Gussone Sicilia), where duplicates by Tineo's specimens are usually hosted, we did not trace any samples but only the unpublished illustration ("Ranunculus schowii Tin./V. Cartoccio dis./1845"; NAP barcode NAP0000512) cited in the protologue. The illustration is also labelled with a representation label [49] "12a. Ranunculus schowii Tin./Aprile, Majo", where "12a" is a reference to the position of the species within the genus in Gussone's Synopsis. We were not able to trace original material in FI, G and P. The unpublished illustration in NAP (NAP0000512) is the only element belonging to the original material (Art. 9.4 of the ICN), agrees with the protologue and is here designated as the lectotype for the name R. schowii (Figure 2). Taxonomic notes: based on the protologue and the lectotype, R. schowii seems to have unique characteristics, only marginally close to particular forms of R. isthmicus Boiss. The individual depicted in the illustration (lectotype) shows fusiform root tubers, basal leaves tripartite with entire or lobed (only at the apex) segments and deflexed sepals at flowering. Further studies to assess the morphological variability in this species and to clarify its taxonomic status will be needed. In the case of synonymy of R. isthmicus Boiss. (published in 1846, [50]) and R. schowii (published in 1845, [18]), the latter would have priority and it should be advisable to proceed with a formal conservation proposal for the name R. isthmicus. Distribution: endemic to the SE Euro-Mediterranean area, distributed in Italy, Balkan Peninsula, Turkey and Georgia. The presence of R. gracilis in the latter country was reported by Grossheim [54], under the name R. agerii Bertol., for a single locality (Akhaltsikhe district; see Map No. 82 included in [54]), but, in our opinion, this report requires further checks. In Italy, in the current state of knowledge, it is present only in Lazio based on our finding reported here, no longer recorded in Emilia-Romagna and Sicily, recorded by mistake in Calabria and Puglia and formerly cultivated in botanical gardens in Toscana and Umbria (Figure 3). The presence of the species in each Italian administrative region is discussed below: • Piemonte: the species was cited as doubtfully occurring in Piemonte by Conti et al. [42] and as recorded by mistake by Bartolucci et al. [4]; it was never recorded for the region (D. Bouvet and A. Selvaggi, pers. comm.). • Emilia-Romagna: the presence of the species is confirmed by several old herbarium specimens kept in BOLO, CHE, FI, G, P, PI and RO (see specimens examined). Furthermore, the 16th century samples preserved in the "En Tibi" and "Cibo" herbaria were also collected in Emilia-Romagna near Bologna (Stefanaki et al. 2018(Stefanaki et al. , 2019. No recent herbarium samples or bibliographic records have been found (see, [55]); therefore, the species should be considered as no longer recorded. Targeted field research will be needed before considering the species as locally extinct. • Toscana: the species was never recorded for the region. We traced an old herbarium specimen in FI, collected in the Botanical Garden of the University of Firenze, where the species was probably cultivated. • Umbria: the species was recorded for the region in the past [9,11,15] as naturalized in the Botanical Garden of the University of Perugia. We traced the herbarium specimen linked to the old report by Fiori [11] in FI (see Specimens examined). • Lazio: during field investigations carried out in the territory of Anagni (Frosinone, Central Italy) in March and April 2022, we discovered the species on Mt. Campitelli ( Figure 3). Our finding corroborates the old report for this area by Sibilia ([56], under the name R. agerii) and confirms the presence of this species in Italy. The Sibilia record has never been incorporated in the regional floras [4,[57][58][59]. In the current state of knowledge, this is the only population present in Italy. • Puglia: the species was recorded by Di Pietro and Misano [60]. This record was later regarded as doubtful by Bartolucci et al. [4]. We were not able to trace herbarium specimens linked to this record and the species should be regarded as probably indicated by mistake (R. Di Pietro, pers. comm.). • Calabria: according to Peruzzi and Passalacqua [41], the Calabrian records should be referred to R. monspeliacus subsp. aspromontanus. • Sicily: the species was reported in Sicily from different localities by Giardina et al. [12]: between Vittoria and Terranova based on the description of R. schowii [18], between Catania and Misterbianco based on Strobl [61] and from Polizzi Generosa [62]. We have shown that R. schowii is not related to R. gracilis; therefore, the report of the latter between Vittoria and Terranova is erroneous. We traced, in PAL, the sample collected in Polizzi Generosa ("sotto il paese di Polizzi Generosa vicino all'acquedotto, 30 April 1990, Raimondo and Certa"), which should be referred to R. paludosus. A specimen cited in Wikiplantbase Sicilia [63] as R. gracilis and stored in PAL (No. 43515) collected at Busambra belongs to R. paludosus as well as a specimen in CAT No. 048272 (Monte Lauro, 9/V/1991, Brullo et al.). The only datum that we were able to confirm is the indication by Strobl [61], for the Amenano between Catania and Misterbianco at the foot of Etna, thanks to the tracing of an old herbarium sample stored in FI, collected in 1874 by Heidenreich at Misterbianco (quoted also by Fiori et al. [10]). Based on our data, R. gracilis should be considered as no longer recorded in Sicily. Targeted field research will be needed before considering the species as locally extinct. Chromosome number: 2n = 16 [64]. Conservation status: Ranunculus gracilis currently occurs outside the NATURA 2000 network on Mt. Campitelli (Anagni, Frosinone) in Lazio (Central Italy). The populations in Emilia-Romagna (Northern Italy) and Sicily, confirmed by old herbarium specimens, have not been observed for over 120 years. The area of occupancy (AOO) is 4 km 2 , calculated with GeoCAT (Geospatial Conservation Assessment Tool) software [65]. The species actually occurs in one location and a decline in the AOO was observed, considering the possible extinction of some populations. According to IUCN [66] criterion B2ab(i,ii,iv), the species is assessed as Critically Endangered (CR) at the regional level (Italy). Conclusions Nomenclature plays a central role in the description of the diversity of life on our planet and the typification process is essential for any taxonomic study. At the same time, floristic research and the study of herbarium collections are of crucial importance in biodiversity conservation of vascular plants and are necessary to collect data for planning the correct conservation strategies. Our study on R. gracilis in Italy allowed us, primarily, to evaluate the taxonomic identity of R. agerii and R. schowii, both described from Italian territory. After typification, R. agerii should be regarded as a heterotypic synonym of R. gracilis, while R. schowii showed a combination of unique characters, close in some ways to atypical forms of R. isthmicus, and needs further studies to assess the morphological variability in the species and to clarify its taxonomic status. Thanks to our contribution, the presence of R. gracilis in Italy is confirmed, expanding the distribution range of this endemic species to the SE Euro-Mediterranean towards the west. In Italy, R. gracilis is present, in the current state of knowledge, with a single population at risk of extinction found in Lazio (Central Italy). Furthermore, we were able to confirm the historical presence of the species, based on the study of herbarium collections, in the Emilia-Romagna (Northern Italy) and Sicily, where it has not been observed for over 120 years. It will now be possible to plan specific field surveys to verify whether R. gracilis is still present in these areas or is to be considered extinct. Furthermore, the species is now reported in the national conservation framework.	2022-11-17T16:16:22.246Z	2022-11-01T00:00:00.000	{ "year": 2022, "sha1": "42f9ab085e98c75127ccd3b96091891c336f8cc8", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2223-7747/11/22/3094/pdf?version=1668430026", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "a2101bece340333413ef21dc0b80467fe175f2ac", "s2fieldsofstudy": [ "Environmental Science", "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
221748735	pes2o/s2orc	v3-fos-license	A Review of Translational Magnetic Resonance Imaging in Human and Rodent Experimental Models of Small Vessel Disease Cerebral small vessel disease (SVD) is a major health burden, yet the pathophysiology remains poorly understood with no effective treatment. Since much of SVD develops silently and insidiously, non-invasive neuroimaging such as MRI is fundamental to detecting and understanding SVD in humans. Several relevant SVD rodent models are established for which MRI can monitor in vivo changes over time prior to histological examination. Here, we critically review the MRI methods pertaining to salient rodent models and evaluate synergies with human SVD MRI methods. We found few relevant publications, but argue there is considerable scope for greater use of MRI in rodent models, and opportunities for harmonisation of the rodent-human methods to increase the translational potential of models to understand SVD in humans. We summarise current MR techniques used in SVD research, provide recommendations and examples and highlight practicalities for use of MRI SVD imaging protocols in pre-selected, relevant rodent models. Electronic supplementary material The online version of this article (10.1007/s12975-020-00843-8) contains supplementary material, which is available to authorised users. Introduction Cerebral small vessel disease (SVD) is estimated to cause 20-25% of strokes globally and 45-65% of dementias [1,2]. Magnetic resonance imaging (MRI) is used extensively in clinics and research to identify SVD-associated lesions and imaging biomarkers. Key SVD-related features, image acquisition and quantification methods are summarised in recent position papers [3,4]. Various rodent models reflect different putative SVD mechanisms and some features of human disease [15][16][17][18][19]. Hypertension models replicate elements of microvessel remodelling from some sporadic human SVD [20], including venous collagenosis [21], but hypertension is only one risk factor. The spontaneously hypertensive stroke prone rat (SHRSP) has endothelial dysfunction, microglial, white matter and BBB abnormalities prior to hypertension and sporadic SVD features when older [18]. Bilateral carotid artery microcoils to induce mild stenosis (BCAS) leads to some SVD characteristics [22,23] but may work through altering carotid elasticity and arterial pulsatility. There are also several monogenic SVD and knockout models [19,24,25]. There are some inherent limitations to the translational potential of animal models [15,16,18]. Anatomically, white:grey matter ratios and brain sizes differ markedly ( Fig. 1) [15,16]; arterial anatomy [26,27], density, spacing and positioning of penetrating arterioles and draining venules also vary [28]. In stroke models, assessing rodent neurological deficits is challenging when symptoms are mild and recovery may be rapid [29]. Anaesthetics, necessary for many types of study, can affect cerebral haemodynamics and CSF transport [30] and may provide neuroprotective or adversive effects that could alter the tissue changes [31,32]. Rodent respiratory and heart rates are higher, restricting options for physiological measurements and pulse gating required for some MR sequences at comparable temporal resolution to humans. Despite the anatomical and physiological differences from humans, rodent SVD models are however key for investigating pathological processes and time trajectories of disease evolution and developing and testing novel therapies [33]. Preclinical MRI facilitates independent validation with contemporaneous histology or other imaging techniques and improves clinical translation and exploration of physiological processes, e.g. fluid flow through the glymphatic system [34][35][36], glymphatic system changes during sleep [37] and effects of risk factors, including hypertension and diabetes, on tissue damage and microvascular fluid dynamics [38,39]. As part of efforts to improve the translational value of preclinical models to human SVD, particularly through the use of MRI, we reviewed the literature to identify studies which had adapted clinical MRI methods to preclinical MRI, or vice versa. Our intention is not to review preclinical SVD rodent models but rather to evaluate synergies, strengths and limitations between human and rodent MRI to optimise the translational potential of MRI for non-invasive longitudinal assessment of disease development and progression in SVD. Systematic Review We reviewed the literature to extract information on approaches to improving comparability or complementarity of brain MRI techniques between studies in rodents and humans. The systematic literature search was conducted on Medline and Embase from 1946 until April 2020 through Ovid. Exploded headings and search terms relating to SVD were combined with terms for MRI and relevant advanced MR techniques. We also combined these results with a comprehensive search strategy for rodents based on Hooijmans et al.'s filter [40] and terms associated with translational research (e.g. translat, retranslat) prior to limiting to papers relating to humans. Finally, we removed duplicates. For the full search strategy, see the Supplementary Material. Fig. 1 Approximate total intracranial volume (TIV) volume (ml) and grey matter:white matter:CSF ratio in healthy (young) animals shown relative to the human brain based on publicly available templates We also manually checked reference lists in reviews and original papers for additional relevant references. Other papers were identified from the authors' libraries. We inspected all identified papers to ascertain whether they satisfied the eligibility criteria. We included papers that provided information on MRI methods designed to be used in humans and those designed for use in rodent models but that aimed to capture SVD features, including static [41] and dynamic biomarkers (e.g. vascular reactivity and BBB dynamics), which use similar sequences in rodents and humans including practical guidance. We excluded studies published solely as conference abstracts due to providing insufficient detail. One author (MSS) extracted summary data for each paper including the type of publication, diseases covered, imaging techniques employed and a short summary of the focus. Other authors resolved uncertainties. We sought to provide an overview of all published strategies for comparable human-rodent MR imaging protocols, practicalities and advice on specific sequence(s) including structural, post-mortem and dynamic vascular function assessments including CVR and DCE. Results The search identified 305 unique publications of which we excluded 260 as irrelevant based on the title, mainly due to being in an unrelated population, case reports or modality (e.g. positron emission tomography (PET), computed tomography (CT)) ( Fig. 2). On full-text review of the remaining 45, a further 30 were excluded, mainly due to the following: no MRI (12), translation of other biomarkers for drug development (five), clinical or preclinical studies only (six), conference abstracts (six) and one book. We found no protocol with guidance on designing longitudinal MRI studies in rodents to mirror typical MRI research to examine disease development in human cohorts. One rodent protocol used DTI, T2w, T1w and T2*w but not FLAIR [22]. No studies addressed image analysis issues that are commonly encountered in human studies, such as image registration and lesion tracking over time, or combining data from several different sequences from the same anatomical regions or lesions at one time point. We did find a few studies where assessment tools developed to assess SVD features in humans, such as the Fazekas scale for WMH [56], Brain Observer MicroBleed Scale (BOMBS) for cerebral microbleeds [57] and several image analysis methods such as segmentation and voxel-based morphometry, have been adapted for use in animal models [22,54]. We also found one example of a novel imaging approach developed to assess perivascular space fluid uptake in rodents that had been translated to human use [53]. There were few studies where MRI had been used to assess response to treatment in rodents [42,43]. The review papers highlighted several complementary methods including microscopy [45] and functional [43,45,46] and in vivo and ex vivo structural MR imaging [22,42,44,45,50] (Table 1). However, these methods were only rarely used together in the same study for validation [49]. Lastly, several technology [44] or transferability [45,46] limitations for clinical to preclinical and vice versa were highlighted, including the need for further validation and methodological advancements to provide scans with higher sensitivity and specificity. There was no detailed overview of a range of MR imaging techniques applied in rodents in disease-or lesion-specific contexts to mirror those developed to study human SVD [3]. Proposed Approaches to Improve the Potential of Rodent-Human Translational MRI Structural MRI in Rodents and Humans Human SVD features are present in many models [15,17,18,50]. Certain features are less commonly reported; however, PVS were only recently identified clinically and appear on histology in pericyte-deficient mice [19]. Therefore, optimised parameters enhancing visibility of disease-related features should be used and standardised where possible (https:// harness-neuroimaging.org/) [4]. Translational SVD research must account for practical differences. Rodent imaging needs ultra-high magnetic field strengths (i.e. ≥ 7 T) as a necessary and common means of increasing signal-to-noise ratio due to smaller spatial resolution. In human imaging, acquisitions can be accelerated via several techniques, including compressed sensing and parallel imaging [59,60], thereby reducing motion artefact while allowing incredibly detailed 3D acquisitions in clinically acceptable times; however, such acceleration methods are often less readily available on preclinical platforms. Variations in preclinical pulse sequences may affect comparability of contrast-to-noise ratio (CNR) and image interpretation (Fig. 3). Applying clinical imaging protocols may better integrate translational work, track disease progression/assessment (Fig. 4) and improve cross-model methodological translatability. Possible exemplar sequences, based on ones found to be reliable and informative at the authors' sites, are listed in Table 2. Options include 2D or 3D sequences, but while 3D are desirable, they take longer than 2D, so 2D may be preferable in some situations. Validation of pathogenic mechanisms, tissue changes and evolution, complemented by invasive measurements, e.g. 2photon [38] and post-mortem microscopy, are main reasons for using preclinical models. Preclinical imaging allows longer/more regular scanning and examination using multiple modalities. Advanced structural scans provide useful additional metrics, e.g. DTI for white matter tract integrity/visualisation, and network connectivity assessment, which may help explore cognitive/functional deficits. Anatomical factors that may affect the translational potential include appropriate metrics/methods to control for relative brain size/tissue ratios and the effects of interventions to increase SVD burden. Clinical SVD features typically increase with age; hence, while using naturally aged rodents [61] may be more challenging and costly, they may reveal more obvious features plus be relevant to longitudinal studies and valuable for preclinical drug testing. [62,63]. Long scans are much more feasible PM yielding better image quality with corresponding benefits to signal-to-noise ratio (SNR), spatial resolution and reduced partial volume, thereby enabling precise MR-histology comparisons. Formalinperfusion fixation improves contrast-to-noise (CNR) for MR microscopy; while SVD lesions remain conspicuous [51,[64][65][66], prolonged fixation may also induce signal artefacts obscuring lesions [67] and discriminating between pre-and post-mortem damage on MRI may be more challenging (e.g. small parenchymal haemorrhage versus post-mortem intravascular thrombus) [68]. While perfusion fixation is the gold standard for preclinical models, immersion is often preferred in humans for practical reasons. Optimal perfusion fixation is also important to avoid post-mortem intravascular thrombus mimicking pre-mortem intravascular thrombus for example [69]. The time of sacrifice and fixation method must therefore be carefully considered. There are validated protocols relating pre-mortem and PM human MR-visible SVD lesions to histology [62,70,71]. While some features may be less evident at PM, PM-MRI is well-suited to automated analysis and is underused in SVD research. Few papers directly compare PM-MRI to histology [72][73][74][75]. Non-quantitative analyses, assessing overall distribution and size [76], risk missing heterogeneous WMH features, including pathological variation [77]. There are limited data on lesion development in rodents during normal ageing; optimisation studies could explore histological-MRI correlates across the lifespan. Close scrutiny of structural and quantitative images by experts in human MRI may identify lesion development stages on PM-MRI. Ex vivo DTI gives data for large areas of tissue and thus can complement and guide the 'spot sampling' approaches typical of histological assessment of white matter integrity [78,79]. Cerebral microinfarcts (CMIs), small ischaemic lesions encompassing neuronal loss, gliosis and cavitation [80], can appear acutely on diffusion imaging. However, size and signal transience may lead to underestimation of frequency and involvement [71,81]. PM-MRI allows direct histopathological validation of specific imaging markers [82] including in rodent models. Studies comparing ex vivo MR with histology of CMIs are sparse. More studies would help determine the role of CMIs in SVD and neurodegeneration and help improve the spatial resolution. Advanced Dynamic MRI Methods In Vivo in Rodents and Humans Cerebral Blood Flow and Cerebrovascular Reactivity CVR refers to vasodilation and vasoconstriction of cerebral microvessels, typically to vasoactive stimuli (e.g. increased CO 2 inspiration or intravenous acetazolamide). CVR is a quintessential cerebrovascular health measure that reflects the role of blood vessels in regulating CBF, oxygen and nutrient delivery, waste product clearance and dissipating heat [93]. CBF is generally measured at rest, with ASL being a widely used method [94]. Several studies have imaged CBF and CVR in SHR and Wistar Kyoto (WKY) rats [55,95]. However, anaesthetic agents can affect resting CBF and haemodynamic responses and therefore are particularly important to consider when planning experiments on CBF or CVR. For example, CBF was higher in SHR versus in WKY under 2% isoflurane but not with alpha-chloralose and isoflurane reduced CVR [96]. As in preclinical functional neuroimaging, sedation rather than full anaesthesia is often preferred [46]. Dexmedetomidine depresses global CBF; therefore, in some experiments, general anaesthesia with inhalational agents may be preferable. Anaesthetic protocols should also minimise impact on signal and derived imaging variables; further optimisation studies and standardised reporting of anaesthetic protocols would be beneficial. CVR requires a physiological manipulation/challenge but voluntary breath holding is unsuited to preclinical studies [97] and controlled hypercapnia via breathing apparatus is more reliable. In rats and larger murine models, intubation provides greatest control of inhaled and exhaled gases and is optimal for preclinical CVR. Sealed chambers are practical where intubation is unviable or lower level sedation is preferred. Since CO 2 tolerance varies between species [98,99], tolerability should be balanced against inducing robust signal changes. Humans tolerate a 6% CO 2 stimulus well [99]. CVR analyses typically use regression models but must account for physiological and practical factors, notably haemodynamic delays, and filtering band frequency may vary under anaesthesia [46]. CVR has not been fully exploited in rodents to better understand how impaired vasoreactivity develops at whole brain level and leads to brain damage in SVD. Longitudinal CVR measurements coupled with multiphoton microscopy via cranial windows or isolated vessel preparations [100] could strengthen the direct validation of in vivo CVR and improve its use as a biomarker and an intermediary outcome in trials of therapeutic interventions. MRI, contrast-enhanced CT (CE-CT), PET and biofluid biomarker approaches can measure BBB permeability [4]. MRI methods include dynamic or static contrast-enhanced MRI [4,11] and non-contrast-based methods (e.g. ASL [108,109] and T1-w black-blood imaging [110]). The most widely used method to detect subtle focal BBB permeability increases is DCE-MRI [102]. Preclinical validation is limited, but in vivo BBB measures would complement histologic assessment and validation of BBB dysfunction. DCE-MRI involves T1-mapping followed by intravenous Gd injection and repeated T1-w sequences [111]. Multi-slice or volume sequences, usually GRE or fast low-angle shot, run repeatedly for ca. 20 min; longer acquisitions may be appropriate for low-level leakage [112,113]. Sequence optimisation balances coverage, SNR and spatial and temporal resolution. As pharmacokinetic analysis requires reliable arterial or venous input functions, identifiable vessels must be covered, e.g. internal carotid arteries or sagittal sinus. High temporal resolution is critical for bolus injections due to rapid blood signal changes [113,114]. A dilute contrast phantom adjacent to the animal's head can allow signal-to-concentration transformation correction. Anaesthesia level, temperature and respiration should be monitored and adjusted to minimise input function variation. BBB function parameters, such as volume transfer constants between extracellular space and plasma, are derived by modelling the Gd concentration-time curve, along with CBF and cerebral blood volume. There is now dedicated software for clinical and preclinical DCE analyses (e.g. ROCKETSHIP [115]). Patlak models are best suited to subtle BBB leakage in SVD [112,113]. Retinal imaging, MR venography [116] and phase contrast MRI [117] have been used as surrogates to direct measurement of vessel diameter in response to stimuli; further studies, particularly in humans at field strengths ≥ 7 T, may provide more detailed insight into properties of the microvasculature. Preclinically, methods like multiphoton microscopy can also complement DCE-MRI by determining the microvascular changes underpinning BBB leakage [118]. There are novel MRI methods in development that use endogenous contrast which may improve sensitivity. For example, DW-pCASL employs diffusion weighting to distinguish labelled blood in the microvasculature from that in brain tissue to measure water exchange [108]. WEPCAST also employs an ASL approach using velocity encoding to isolate venous signal [109]. However, these approaches are hypothesis-based and their translation to humans would be greatly facilitated by first demonstrating that they provide reliable measures of BBB function in preclinical models. MR Spectroscopy to Assess Metabolites Magnetic resonance spectroscopy (MRS) determines tissue metabolite levels in vivo and would provide a promising approach to explore SVD and neurodegenerative disease progression [14,[119][120][121][122]. However, MRS has not been applied extensively in human SVD or preclinical models. There are some practical limitations. For example, for sampling homogeneous tissue, single voxel spectroscopy (SVS) in mice requires typical volumes ca. 0.008 cm 3 [123] (0.4% of brain volume) relative to 4 cm 3 (0.03%) in humans [124] and positioning the sample volume requires structural imaging. Multi-voxel MR spectroscopic imaging (MRSI) [125] increases brain coverage and can examine metabolite distribution, although with longer acquisition times. It is important to control for disease burden and relative proportions of healthy/diseased tissue, particularly in advanced disease, and where atrophy reduces the amount of tissue to sample. It is important to establish the test-retest reliability on individual MRI scanners, particularly for lactate and coupled metabolites [126] before use in experiments although the reproducibility for detecting more abundant metabolites is generally good and several quantification approaches are available [124]. Surface and/or refined coil designs improve rodent MRS sensitivity [127] and higher field strengths provide better spectral resolution and distinction of metabolite peaks. Preclinical MRS allows cross-validation of metabolite concentrations using chemical methods, providing greater confidence scanning humans longitudinally. Beyond proton MRS, carbon-13, oxygen-17, sodium-23 and phosphorus-31 MRS may be applied, though multinuclear equipment is needed. Contrast agents, notably deuterium [128] and thulium [129], can monitor metabolism, pH and temperature in vivo to assess disease progression or changes in ketogenic states. Novel MRI Methods in Rodents and Scope for Rodent-Human Translation PVS are small conduits that envelop penetrating cerebral arterioles/venules where CSF can exchange with interstitial fluid (ISF) [48]. As part of the glymphatic system, PVS are thought to clear brain fluid and waste, facilitated by aquaporin-4 (AQP4) water channel-mediated CSF-ISF exchange at the peri-capillary space before clearance to lymphatic vessels [34,130,131]. While aspects of the physiology are controversial [132,133], CSF-ISF exchange studies provide opportunities to understand PVS in vascular and neurodegenerative diseases. PVS become enlarged and more visible in SVD and are associated with inflammation, impaired CVR, increased BBB permeability and vascular pulsatility [48]. Their small size makes them difficult to assess in humans. However, CSF delivery to PVS can be characterised in rodents using DCE-MRI and Gd injection into the cisterna magna CSF. Small volume infusion of Gd into the CSF pool during 3D MRI demonstrates spatially and temporally resolved solute transport through the brain [134] and can show altered PVS function with vascular risk factors. For example, type 2 diabetes mellitus rats exhibit slower Gd clearance, with accumulation and retention, enhanced perivascular arterial influx and increased hippocampal signal intensity [39]. Tracer transport mechanisms are highly complex, but pharmacokinetic models [135,136] and mass transport algorithms [137,138] help quantify influx/efflux contributions. An optimal mass transport analysis in SHRSP rats reveals reduced and slowed solute transport from CSF into the brain [139]. Though rare, opportunistic human studies show CSF solute transport into basal brain parenchyma over longer times with similar distributions [140][141][142]. Gd injection into the cisterna magna in humans is not a practical technique; Gd injections into the lumbar CSF and tracking through the intracranial CSF have been done rarely and only when diagnosing pathological conditions. More clinically applicable although less sensitive techniques to track PVS function include diffusion imaging, PC-MRI and ultrafast MR imaging. In rodents, T2-w diffusion techniques showed CSF in PVS preferentially moved parallel to blood flow fluctuating with cardiac pulsation, consistent with PVS-CSF movement [143] but has yet to be applied in humans. PC-MRI shows that intracranial arterial, venous and CSF pulsatility in the main cisterns, correlate with WMH burden in SVD [144][145][146], depends on directional flow and is less suited to understanding water mobility within the brain. Magnetic resonance encephalography (MREG) is an emerging high temporal resolution sequence, which is thought to reveal spatial-temporal patterns driven by different cardiac, respiratory and vasomotor forces highlighting that cerebral water movement is directional and cyclic with several drivers [147]. Recent work with APQ4-deficient mice suggests multi-echo ASL may provide insight into clearance mechanisms [148]. Such methods may reveal new insights into SVD, particularly large calibre vessel pulsatility effects on spatiotemporal water movement characteristics, although preclinical validation remains key. Discussion Major advances have occurred in understanding human SVD, thanks to modern MRI methods; however, large gaps in knowledge remain which could be addressed through a range of SVD models and capitalising on multiple forms of clinically relevant image contrasts available with MRI. We demonstrate the need for greater transferability and reproducibility of preclinical-clinical MRI findings. There are numerous relevant rodent models for SVD. However, as yet, the imaging approaches do not appear to be taking full advantage of the knowledge derived from characterising human disease and thus limit the translational potential of rodent models in SVD. Further studies to ascertain key features of SVD and disease progression would help focus preclinical rodent models on the most salient features, whether structural or dynamic measures (e.g. WMH and PVS burden, cerebrovascular reactivity, BBB leakage etc). Many of the imaging techniques commonly used clinically have already been applied to various rodent models, including structural MR and techniques to investigate BBB integrity. However, several structural sequences are necessary in human MRI to capture the features properly and this approach could improve the yield of preclinical MRI. Closer matching of clinical to preclinical imaging protocols may aid comparisons of data and provide a fuller picture of differences in SVD disease progression and manifestations. Only limited use has been made of ex vivo MRI and histology in studying SVD to date, though potentially relevant protocols exist for several relevant features. Greater use of these techniques may help determine which cellular mechanisms are at work, while improving understanding of the genesis and evolution of lesions and other imaging features, such as WMH and CMBs. Availability of human tissue, particularly from intermediate disease states, remains an obstacle; however, many questions can be explored with greater use of animal histology, in vivo and PM-MRI. For dynamic imaging methods in rodents, determining the optimal anaesthetic regimen and the effect on derived imaging metrics is a significant challenge, but would help improve comparability between preclinical and hence clinical studies. Rodent models allow direct in vivo validation which would significantly advance understanding of underlying disease mechanisms, and emerging imaging methods, including synthetic MRI [149], with the potential to quantify disease-related tissue properties in vivo. Using comparable processing and analysis methods in preclinical and clinical imaging will greatly increase translational potential [113]. Application of similar image analysis methods to animal and human studies is entirely feasible, would avoid repeating the same errors, improve translation and is encouraged [22,54,115]. Preclinical models have also enabled advances in understanding of microvascular dysfunction underlying SVD and methods to measure paravascular transport. While the initial approaches, based on relatively invasive techniques, are not applicable in routine clinical studies, it has stimulated interest in alternative imaging methods for humans which show some promise. Imaging of SVD is key to advancing understanding of disease pathophysiology and aiding the development of novel treatments. There is immense untapped potential for clinical research to inform preclinical work and vice versa. To maximise the benefits of research into SVD, there is a need for greater engagement and active collaborations between clinical and preclinical researchers to develop research programmes taking full account of the latest advances in both domains. Compliance with Ethical Standards Conflict of Interest The authors declare that they have no conflict of interest. Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed Consent Informed consent was obtained from all individual participants included in the study. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.	2020-09-17T13:06:19.258Z	2020-09-16T00:00:00.000	{ "year": 2020, "sha1": "52b0a01dd30e8c2755aef2f565d58b7bf4048d8e", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/s12975-020-00843-8.pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "e091a59ef481b3cca356a7e54772b7286a34027b", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
253582889	pes2o/s2orc	v3-fos-license	Genome sequence of SARS-CoV-2 sublineage BA.2.12.1 strain from Bangladesh Arifa Akram National Institute of laboratory Medicine and Referral center, Dhaka Syed Muktadir Al Sium Bangladesh Council of Scientific and Industrial Research (BCSIR), Dhaka Toufiqul Islam Novus Clinical Research Services Limited, Bangladesh Md. Ashiqur Rahman Novus Clinical Research Services Limited, Bangladesh MD. Mahbub Ali Novus Clinical Research Services Limited, Bangladesh Md. Raihanul Islam Novus Clinical Research Services Limited, Bangladesh Uttom Kumar Bhowmik Novus Clinical Research Services Limited, Bangladesh Md Alimur Reza Beximco Pharmaceuticals Limited, Dhaka, Bangladesh Dear Editor Over the last few months, multiple lineages of the omicron (B.1.1.529) variant of SARS CoV-2 have emerged. BA.1, BA.2, BA.3, BA.4 and BA.5 are the first five branches descending from an original Omicron ancestor. Recently the United States first detected Omicron sub-variant BA.2.12.1 which is the 12th lineage to branch off from BA.2 [1]. The Omicron variant contains more than 30 mutations and most of the mutations were reported in the spike protein which has been used as a key target for most of the available vaccines. Omicron variant is three times more infectious than the original SARS-CoV-2 strain [2]. It is the seventh month of 2022 and Bangladesh has already faced its record highest positivity of SARS-CoV-2 early in this calendar year. In recent days, BA.4, BA.5 and BA.5.1 along with BA.2.12.1 lineage of Omicron continues the emergence and evolution of a new SARS-CoV-2 epidemic Globally [3] along with South Asian countries [4] and in Bangladesh. On 28th January 2022 the positivity rate of SARS-CoV-2 infection was 33% which was the highest since 8th March 2020 when the first case was reported in Bangladesh and declined to below 1% on 16th March 2022 [4]. Though all the prohibiting orders have been revoked after the Delta outbreak at the end of 2021 as combatting SARS-CoV-2 has become a new normal, all the government and private service activities along with educational activities in institutes continue to be operative. Since then, the positivity rate maintained between 0-1% until 10th June 2022. All of a sudden the positivity rate tempted to surge forward. After 10th June 2022, a sharp incline of SARS-CoV-2 positivity rate has been noticed, which has raised to 16% on 26th June 2022 just in a difference of 15 days. To determine the viral lineage of SARS-CoV-2, we performed whole genome sequencing analyses using SARS CoV-2positive samples collected at Novus Molecular lab, Poribag, Dhaka and sequened at Child Health Research Foundation (CHRF), Dhaka. The Sequencing library was prepared using Illumina COVIDSeq Assay kit and sequencing was performed in Illumina iSeq100 platform. Its prevalence in Bangladesh should be monitored closely because BA.2.12.1 can escape antibodies elicited by previous Omicron infection even after having been vaccinated with booster dose. World health organization emphasized to monitor BA.2.12.1 sub variants closely on May 4, 2022 [5]. Omicron BA.2.12.1 sub variants spread very fast, which led to the resurgence of the epidemic in many parts of the united states and cases have been reported in at least 79 countries all over the world. In the current situation, there is very few evidence of the tenable threats associated with the emergence of the newly Omicron BA.2 lineage. As already countries along with many South Asian countries emerged by newer Omicron sub variant, Bangladesh is in an alarming period of the new wave [3]. A pick high infectivity could lead to an even more destructive situation over the countries system and it will be tough to control and providecare as more people will get infected/sick. Rapid screening and monitoring of newly coming BA.2.12.1 lineage of SARS-CoV-2 and genomic surveillance is essential for early detection of any other emerging variants to take proper measures of prevention. Evolution of next generation vaccines could be the effective prevention of concurrently evolving and emerging newer SARS-CoV-2 variants and lineages. It is the foremost sub variant of Omicron to be conscious of the evolutionary process at present. Though this is the first reported BA.2.12.1 from Bangladesh, epidemiologists must investigate to monitor their prevalence (see Fig. 1). Conflicts of interest There are no conflicts of interest. Data extraction: Md. Ashiqur Rahman, Md Alimur Reza. All authors contributed to the reviewing for important intellectual context and approved of the manuscript to submit. Conflicts of interest There are no conflicts of interest.	2022-11-18T14:06:13.192Z	2022-11-01T00:00:00.000	{ "year": 2022, "sha1": "4c418e7f188dd29048884fdbbcfcc933e2e47ff2", "oa_license": "CCBYNCND", "oa_url": "https://doi.org/10.1016/j.nmni.2022.101050", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "dc50ffd07274874094bf1d8893a4c17d6350490a", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
268402981	pes2o/s2orc	v3-fos-license	CoII-catalysed synthesis of N-(4-methoxyphenyl)-5-(pyridin-4-yl)-1,3,4-oxadiazol-2-amine hemihydrochloride monohydrate The CoII-catalysed synthesis and crystal structure is reported for the title compound, which features a symmetric N⋯H+⋯N unit. Chemical context 1,3,4-Oxadiazole derivatives have been studied in recent years for their diverse biological activities (Gond et al., 2023;Abd-Ellah et al., 2017;Bitla et al., 2020).As a result of their electron-accepting properties, high quantum yield, and good thermal and chemical stabilities, they have also been used in electroluminescent, optical and electron-transporting materials and chelating agents (Najare et al., 2020;Wu et al., 2012).Several methods for the synthesis of 1,3,4-oxadiazoles from acyclic precursors are available, which include oxidative cyclization of acylhydrazones (Jedlovska ´& Les ˇko, 1994) and acylthiosemicarbazides (Omar et al., 1996, Paswan et al., 2015).In the presence of a strong acid, an N-acylhydrazine carbodithioate is converted into a thiadiazole whereas in the presence of a weak acid or base or on complexation they can be cyclized into oxadiazole (Reid & Heindel, 1976;Jasinski et al., 2011). We have previously reported the cyclo-desulfurization of several N-acylhydrazine carbodithioates into the corresponding 1,3,4-oxadiazole in the presence of manganese(II) acetate via the loss of H 2 S where the Mn II ion presumably behaves as a weak Lewis acid (Paswan et al., 2015(Paswan et al., , 2016;;Gond et al., 2022).In the present work, a similar reaction is reported in presence of Co II chloride.Similar Co II -assisted cyclization reactions are also reported in the literature (Li et al., 2021(Li et al., , 2023;;Bharty et al., 2012). Supramolecular features In the extended structure, two organic molecules are linked through their pyridine nitrogen atoms via the proton of the hydrochloric acid, which lies on a crystallographic twofold axis.This strong, symmetrical, almost linear N4� � �H4N� � �N4 hydrogen bond (Table 1) leads to a rod-like dimeric structure.These units form a layer-like structure when viewed along b axis of the unit cell (Fig. 3).The water molecules and chloride ions (site symmetry 2) are embedded in the space between the chains and are connected to them via N-H� � �O and O-H� � �Cl hydrogen bonds, thereby generating [001] chains.Weak C-H� � �Cl interactions are also observed (Table 1; Fig. 2). Hirshfeld Surface Analysis To gain further insight into the intermolecular interactions, a Hirshfeld surface analysis was performed using Crystal The molecular structure of the title compound showing 30% probability displacement ellipsoids with hydrogen bonds indicated by dashed lines. Figure 2 The packing of the title compound viewed along the a-axis direction. Figure 3 The packing of title compound viewed along the b-axis direction. Explorer 17.5 (Spackman et al., 2021).Fig. 4a,b shows the Hirshfeld surface mapped over d norm .The red spots show the various hydrogen bonds noted above. Synthesis and crystallization 2-Isonicotinoyl-N-(4-methoxyphenyl)hydrazine-1-carbothioamide was prepared by adding 1.652 g (10.00 mmol) of 4methoxy phenyl isothiocyanate in ethanol solution to 1.370 g (10.00 mmol) of isonicotinohydrazide and the reaction mixture was refluxed for 6 h at 333 K. Upon cooling, a white precipitate of 2-isonicotinoyl-N-(4-methoxyphenyl)hydrazine-1-carbothioamide was obtained (Fig. 5), which was filtered off and washed with a 50:50 v/v mixture of water and ether.Then, 1.00 mmol of 2-isonicotinoyl-N-(4-methoxyphenyl)hydrazine- The IR spectrum (KBr disc) shows an absorption band at 3280 cm À 1 due to the NH group.The C O band is absent and a new band is observed at 1623 cm À 1 corresponding to the C N bond.In addition, a blue shift is observed for the N N band at 1179 cm À 1 compared to the single bond in the thiosemicarbazide intermediate (Fig. 1 in the supporting information).All these data indicate that the carbothioamide moiety has been transformed into the corresponding oxadiazole (Chandra et al., 2022;Jaiswal et al., 2023ab). The 1 H NMR spectrum of the title compound in DMSO-d6 displays peaks at � 10.69 ppm due to the NH proton, at � 8.91 and 7.90 ppm due to the pyridyl ring protons and at � 7.55 and 6.98 ppm due to phenyl ring protons.The methoxy protons appear at � 3.74 ppm.(Fig. 2 in the supporting information).In the 13 C NMR spectrum, peaks at � 156.9 and 155.2 ppm arise from oxadiazole ring carbon atoms, the methoxy C atom appears at 55.7 ppm and the phenyl and pyridyl carbon atoms are observed in the range � 114.8-132.4ppm (Fig. 3 in the supporting information).An absorption at 338 nm in the electronic spectrum of the title compound can be attributed to its �-�* transition (Fig. 4 in the supporting information).It displays fluorescence at 418 nm upon excitation at 338 nm (Fig. 5 in the supporting information) when dissolved in 10 À 5 M DMSO solution. Figure 5 Synthesis scheme for the title compound. Refinement Crystal data, data collection and structure refinement details are summarized in Table 2. Atom H4N was freely refined.Other H atoms were placed in idealized locations (N-H = 0.86 A ˚, C-H = 0.93-0.96 A ˚) and refined using a riding model with U iso (H) =1.2U eq (C,N) or 1.5U eq (C-methyl).Asymmetric N-H� � �N/N� � �H-N refinements with the H atom displaced towards one of the N atoms were inconclusive and atom H4N was placed on the twofold axis. Special details Geometry.All esds (except the esd in the dihedral angle between two l.s.planes) are estimated using the full covariance matrix.The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry.An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s.planes. Figure 1 Figure 1 1-carbothioamide was dissolved in a 50:50 v/v mixture of methanol and chloroform, and a methanolic solution of 0.5 mmol of CoCl 2 •6H 2 O was added and stirred for 2 h, during which time the smell of H 2 S was noted.The clear solution obtained was kept for crystallization and after 15 days, palepink blocks of the title compound were grown.Yield: 60.6%; m.p. 495-498 K. Analysis calculated for C 14 H 12 N 4 O 2. 0.5 HCl•H 2 O: C, 55.21; H, 4.79; N, 18.39%; found: C, 55.25; H, 4.50; N, 18.55%. Table 2 Experimental details.	2024-03-15T15:59:10.852Z	2024-03-01T00:00:00.000	{ "year": 2024, "sha1": "ed0cda97929546381fac89203a04ff9380819655", "oa_license": "CCBY", "oa_url": "https://journals.iucr.org/e/issues/2024/04/00/hb8088/hb8088.pdf", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "726fa6fa9881251f15d92f181ba3d03e40509a51", "s2fieldsofstudy": [ "Chemistry" ], "extfieldsofstudy": [] }
249262638	pes2o/s2orc	v3-fos-license	A Review on Vitis vinifera L.: The Grape Vitis vinifera , the Grape being a common fruit with great nutritional values, has been known to mankind since ancient times. Vitis vinifera is a renowned species of grape with a number of varieties originated from western Asia and southern Europe. Vitis vinifera is the Latin name applied to grapevines. It is a vigorous, high-powered tendril climber with a large, lobed, bright green leaves. The tiny green summer flowers are followed by late summer bunches of small grapes. The fruit is used as food supplement and the seeds and leaves are employed in herbal therapy. The most significant application of grape is in wine production followed by raisin and juice. Thus, grape is also an economically important fruit crop in the world. Several varieties of Vitis vinifera are available in India. Phytochemistry, pharmacological, nutraceutical, traditional uses of Vitis vinifera are been presented in this review. The nutritional and phytochemical constituents present in the grape have resulted in its health beneficial effects however more studies are needed regarding the genotoxicity and toxicity of Vitis vinifera . INTRODUCTION itis vinifera L. commonly known as grape, belongs to the family Vitaceae. Grape is one of the largest commodities in agriculture. The grape farming is called as viticulture. Around 10,000 varieties of grape are there in this world. The varieties include seedless, nonseedless and also come in white, red, green colors. Vitis vinifera species dominate the other species of grape by 90% 1 . Grapes have been used for thousand years because of their nutritional and medicinal benefits. These are rich in sugars, flavonoids, anthocyanin sand proanthocyanins, organic acids, tannin, mineral salts and vitamins 2 . Grapes have been used traditionally in Pakistan, Italy and Turkey as laxatives, carminatives and as drug therapy for cold, flu, anaemia, wound care, allergies and bronchitis. Researches has proven that the bioactive compounds present in grapes has led to the pharmacological activities such as antioxidant, antidiabetic, anticancer, antiinflammatory, anti-acne, anti-aging, antiplatelet, antiasthma, anti-obesity and anti-sunburn and wound healing properties 1 . Active Constituents According to the recent studies, the consumption of grape and grape products has shown the beneficial health effects which is attributed by the unique mix of polyphenolic compounds 5 . Fruit: A good source of bio flavonoids (Vitamin P), malic acid and tannic acid; dehydroascorbic acid, cholesterol, ergosterol and beta-sitosterol. Polyphenols: Grapes are rich in polyphenols and about 60-70% of grape polyphenols are present in grape seeds. Grape seeds also contain procyanidins and proanthocyanidins. These are esters of gallic acid. Anti-inflammatory action The studies have shown that the grape polyphenols decrease the chronic inflammation by modifying the inflammatory pathways or by reducing the ROS levels. Flavonoids and proanthocyanidins present in grapes target multiple pathways to overcome chronic inflammation thus proven to be more effective than the synthetic monotargeted anti-inflammatory drugs 5,6 . Proanthocyanidins extracted form grape seeds found to have immunemodulatory role in inflammatory condition caused by overproduction of nitic oxide and prostaglandin E2 7 . Anti-oxidant activities The consumption of dietary flavonoids extracted from the grape in the form of grape extract and grape seed powder have shown to be effective in suppressing the oxidative stress and preventing the oxidative damage in vivo. These activities of grape are attributed by the functions of grape flavonoids as free radical scavengers and metal chelating compounds 5 . Anti-microbial activity The plant polyphenols have demonstrated to have activities such as antimicrobial, antifungal and antiviral. The different parts of Vitis vinifera with phenolic compounds showed different anti-microbial properties. According to the researches the grape seed extracts are more anti-microbial than the other parts of the grapes. The increasing order of grape anti-microbial activity is from flesh, whole fruit grape extract, fermented pomace, skin, leaves and seed 8 . Resveratrol, a phenolic compound in grape have shown to possess antifungal activity against the human pathogenic fungi Candida albicans and the notable benefit of polyphenols against the chemical derived drug was that there was no induction of haemolytic activity on human erythrocytes. Thus, the observed anti-fungal activity of grapes has been attributed to their commercial applications and are being incorporated into the skincare cosmetics 5,8 . Hepatoprotective activity The polyphenols present in the grape has led to hepatoprotective activity due to their anti-inflammatory and antioxidant properties 9 .It was found that the polyphenol rich grape skin extract improved liver steatosis and protective against diet induced adiposity and hepatic steatosis. The effect is probably because of the suppression of lipogenic enzymes in liver and adipose tissues and modulation of lipid metabolism by regulation of mRNA expression of enzymes, involved in regulation of lipogenesis and fatty acids oxidation 10 . Cardiovascular benefits Numerous researches suggested that the daily intake of grape and/or grape products might protect the cardiovascular health. This protective activity is due to enhanced vascular and endothelial function, decreased oxidation of low-density lipoprotein (LDL), positive alteration in blood lipid concentrations and modulating inflammatory process 9,11 . Anti-cancer activity Some studies suggest that the consumption of grape components could be associated with reduced risk of certain cancers such as colon cancer, breast cancer etc. Grape antioxidants play a major role in their anticancer activity because of their antioxidant, anti-inflammatory and anti-proliferative properties. Antioxidants present in the grape have shown to induce cell cycle arrest and apoptosis in the cancer cells and also prevents carcinogenesis and cancer progression in study models. The mechanism of anti-cancer action is due their effect on multiple cellular events associated with tumour initiation, promotion and progression 12 . Anti-obesity and anti-diabetic activity Polyphenols present in the grape and grape products are suggested to be effective in reducing the metabolic syndrome and preventing the obesity and type 2 diabetes by their action as multi-target modulators with antioxidant and anti-inflammatory effects 5 . Dermatological effects Polyphenols present in the red grape seed extract was found to be having protective effect against multiple doses of UV-B irradiation and also showed enhanced anti-oxidant activity against UV-B irradiation and also inhibits apoptosis due to irradiation to certain extent. Thus, grape extract can be used as a portion in sunscreen 9 . Grape as nutraceuticals Wine being a most widely popular as well as nutritional grape product has been proven to be having beneficial health effects on the human body. The consumption of red wine in moderate amount in daily diet considered to contribute to the consumers health mainly because of their composition of quercetin and resveratrol. However, since alcohol is also its composition its mass consumption is restricted 13 . Unique combination of active constituents in grape like polyphenols, flavonoids, anthocyanins, proanthocyanins, stilbenes, has resulted in development of novel nutraceutical products. There are wide range of food additives and nutraceutical products originating from grape in the worldwide market. Some of the examples for commercialized products are grape skin extract, seed extract, grape skin powder, dry seed powder, pomace powder, anthocyanin colorants etc 5 . Anti-aging activity Pollution and radiation cause the aging of skin. According to some researches grape seed extract could slow the onset of skin aging or may also work by increasing the skin elasticity because of its resveratrol content and also because of its richness in anti-oxidants, thus making it a potential ingredient in anti-aging products 1 . Anti-acne activity The most common skin problem, acne vulgaris is caused by Propionibacterium acnes. Based on some researches the grape leaves extract was found to have anti-acne activity against P. acnes 14 . Traditional uses of Vitis vinifera Fruits of Vitis vinifera is been in use since ancient times for its nutritional benefits and as food. The ripe fruit is been used traditionally as laxative, purgative, fattening, diuretic, aphrodisiac, appetizer and was also used as a remedy for asthma, jaundice, strangury and some blood diseases. The ashes of stem are used as pain reliver for joint pain, swelling of testicles and for piles. The flowers of grapes are been used as expectorant, haematinic and also useful for bronchitis. In Iran, leaves of Vitis vinifera are used in traditional food and also for the treatment of diarrhoea and bleeding. Sap of the young branches of grapes are used as remedy for skin diseases. Dried fruits are in as demulcent, cooling, sweet, laxative, stomachic, used in thirst, heat of body, cough. In folk remedy, the malagma of seed is been used for condylomata of joints. Juice of grape prepared in various manners are been used traditionally to treat various tumours 2 . CONCLUSION Grape and grape-products should be promoted in our daily diet, not only as a heathy food but also as a nutrient because of its varied health beneficial constituents. A number of studies have strongly suggested the inclusion of grape and grape-products as a health-supplement in the daily diet for its significant health benefits. Even though Grape has been in use since ancient times in folk medicine as a remedy for several disease conditions, yet it hasn't been clearly documented. A thorough study needs to be conducted to prove its significance. Active constituents present in grapes have resulted in its potential pharmacological activities but further research has to be carried out to evaluate other possible health and toxic effects.	2022-06-02T15:29:58.267Z	2022-05-15T00:00:00.000	{ "year": 2022, "sha1": "636d42755f696234fd9d56762002f34683c86b83", "oa_license": null, "oa_url": "https://doi.org/10.47583/ijpsrr.2022.v74i01.023", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "d81b30520d8e7bf504fac3b097de8158a8cb1764", "s2fieldsofstudy": [ "Agricultural and Food Sciences" ], "extfieldsofstudy": [] }
11879675	pes2o/s2orc	v3-fos-license	Multi-Metric Optimization Using Ensemble Tuning This paper examines tuning for statistical machine translation (SMT) with respect to multiple evaluation metrics. We propose several novel methods for tuning towards multiple objectives, including some based on ensemble decoding methods. Pareto-optimality is a nat-ural way to think about multi-metric optimization (MMO) and our methods can effectively combine several Pareto-optimal solutions, obviating the need to choose one. Our best performing ensemble tuning method is a new algorithm for multi-metric optimization that searches for Pareto-optimal ensemble models. We study the effectiveness of our methods through experiments on multiple as well as single reference(s) datasets. Our experiments show simultaneous gains across several metrics (BLEU, RIBES), without any signiﬁcant reduction in other metrics. This contrasts the traditional tuning where gains are usually limited to a single metric. Our human evaluation results conﬁrm that in order to produce better MT output, optimizing multiple metrics is better than optimizing only one. Introduction Tuning algorithms are used to find the weights for a statistical machine translation (MT) model by minimizing error with respect to a single MT evaluation metric. The tuning process improves the performance of an SMT system as measured by this metric; with BLEU (Papineni et al., 2002) being the most popular choice. Minimum error-rate training (MERT) (Och, 2003) was the first approach in MT to directly optimize an evaluation metric. Several alternatives now exist: MIRA (Watanabe et al., 2007;Chiang et al., 2008), PRO (Hopkins and May, 2011), linear regression (Bazrafshan et al., 2012) and ORO (Watanabe, 2012) among others. However these approaches optimize towards the best score as reported by a single evaluation metric. MT system developers typically use BLEU and ignore all the other metrics. This is done despite the fact that other metrics model wide-ranging aspects of translation: from measuring the translation edit rate (TER) in matching a translation output to a human reference (Snover et al., 2006), to capturing lexical choices in translation as in METEOR (Lavie and Denkowski, 2009) to modelling semantic similarity through textual entailment (Padó et al., 2009) to RIBES, an evaluation metric that pays attention to long-distance reordering (Isozaki et al., 2010). While some of these metrics such as TER, ME-TEOR are gaining prominence, BLEU enjoys the status of being the de facto standard tuning metric as it is often claimed and sometimes observed that optimizing with BLEU produces better translations than other metrics (Callison-Burch et al., 2011). The gains obtained by the MT system tuned on a particular metric do not improve performance as measured under other metrics (Cer et al., 2010), suggesting that over-fitting to a specific metric might happen without improvements in translation quality. In this paper we propose a new tuning framework for jointly optimizing multiple evaluation metrics. Pareto-optimality is a natural way to think about multi-metric optimization and multi-metric optimization (MMO) was recently explored using the notion of Pareto optimality in the Pareto-based Multi-objective Optimization (PMO) approach (Duh et al., 2012). PMO provides several equivalent solutions (parameter weights) having different trade-offs between the different MT metrics. In (Duh et al., 2012) the choice of which option to use rests with the MT system developer and in that sense their approach is an a posteriori method to specify the preference (Marler and Arora, 2004). In contrast to this, our tuning framework provides a principled way of using the Pareto optimal options using ensemble decoding . We also introduce a novel method of ensemble tuning for jointly tuning multiple MT evaluation metrics and further combine this with the PMO ap-proach (Duh et al., 2012). We also introduce three other approaches for multi-metric tuning and compare their performance to the ensemble tuning. Our experiments yield the highest metric scores across many different metrics (that are being optimized), something that has not been possible until now. Our ensemble tuning method over multiple metrics produced superior translations than single metric tuning as measured by a post-editing task. HTER (Snover et al., 2006) scores in our human evaluation confirm that multi-metric optimization can lead to better MT output. Related Work In grammar induction and parsing (Spitkovsky et al., 2011;Hall et al., 2011;Auli and Lopez, 2011) have proposed multi-objective methods based on roundrobin iteration of single objective optimizations. Research in SMT parameter tuning has seen a surge of interest recently, including online/batch learning (Watanabe, 2012;Cherry and Foster, 2012), large-scale training (Simianer et al., 2012;He and Deng, 2012), and new discriminative objectives (Gimpel and Smith, 2012;Zheng et al., 2012;Bazrafshan et al., 2012). However, few works have investigated the multi-metric tuning problem in depth. Linear combination of BLEU and TER is reported in (Zaidan, 2009;Dyer et al., 2009;Servan and Schwenk, 2011); an alternative is to optimize on BLEU with MERT while enforcing that TER does not degrade per iteration (He and Way, 2009). Studies on metric tunability (Liu et al., 2011;Callison-Burch et al., 2011;Chen et al., 2012) have found that the metric used for evaluation may not be the best metric used for tuning. For instance, (Mauser et al., 2008;Cer et al., 2010) report that tuning on linear combinations of BLEU-TER is more robust than a single metric like WER. The approach in (Devlin and Matsoukas, 2012) modifies the optimization function to include traits such as output length so that the hypotheses produced by the decoder have maximal score according to one metric (BLEU) but are subject to an output length constraint, e.g. that the output is 5% shorter. This is done by rescoring an N-best list (forest) for the metric combined with each trait condition and then the different trait hypothesis are combined using a system combination step. The traits are in-dependent of the reference (while tuning). In contrast, our method is able to combine multiple metrics (each of which compares to the reference) during the tuning step and we do not depend on N-best list (or forest) rescoring or system combination. Duh et. al. (2012) proposed a Pareto-based approach to SMT multi-metric tuning, where the linear combination weights do not need to be known in advance. This is advantageous because the optimal weighting may not be known in advance. However, the notion of Pareto optimality implies that multiple "best" solutions may exist, so the MT system developer may be forced to make a choice after tuning. These approaches require the MT system developer to make a choice either before tuning (e.g. in terms of linear combination weights) or afterwards (e.g. the Pareto approach). Our method here is different in that we do not require any choice. We use ensemble decoding ) (see sec 3) to combine the different solutions resulting from the multi-metric optimization, providing an elegant solution for deployment. We extend this idea further and introduce ensemble tuning, where the metrics have separate set of weights. The tuning process alternates between ensemble decoding and the update step where the weights for each metric are optimized separately followed by joint update of metric (meta) weights. Ensemble Decoding We now briefly review ensemble decoding which is used as a component in the algorithms we present. The prevalent model of statistical MT is a log-linear framework using a vector of feature functions φ: The idea of ensemble decoding is to combine several models dynamically at decode time. Given multiple models, the scores are combined for each partial hypothesis across the different models during decoding using a user-defined mixture operation ⊗. (2) propose several mixture operations, such as log-wsum (simple linear mixture), wsum (log-linear mixture) and max (choose lo-cally best model) among others. The different mixture operations allows the user to encode the beliefs about the relative strengths of the models. It has been applied successfully for domain adaptation setting and shown to perform better approaches that pre-compute linear mixtures of different models. Multi-Metric Optimization In statistical MT, the multi-metric optimization problem can be expressed as: where where N (f ; w) is the decoding function generating a set of candidate hypotheses H based on the model parameters w, for the source sentences f . For each source sentence f i ∈ f there is a set of candidate hypotheses {h i } ∈ H. The goal of the optimization is to find the weights that maximize the function g(.) parameterized by different evaluation metrics M 1 , . . . , M k . For the Pareto-optimal based approach such as PMO (Duh et al., 2012), we can replace g(·) above with g PMO (·) which returns the points in the Pareto frontier. Alternately a weighted averaging function g wavg (·) would result in a linear combination of the metrics being considered, where the tuning method would maximize the joint metric. This is similar to the (TER-BLEU)/2 optimization (Cer et al., 2010;Servan and Schwenk, 2011). We introduce four methods based on the above formulation and each method uses a different type of g(·) function for combining different metrics and we compare experimentally with existing methods. PMO Ensemble PMO (Duh et al., 2012) seeks to maximize the number of points in the Pareto frontier of the metrics considered. The inner routine of the PMO-PRO tuning is described in Algorithm 1. This routine is contained within an outer loop that iterates for a fixed number iterations of decoding the tuning set and optimizing the weights. The tuning process with PMO-PRO is independently repeated with different set of weights for metrics 1 yielding a set of equivalent solutions if h ∈ F then add (1, h) to T 9: else add ( , h) to T (see footnote 1) 10: w p ← PRO(T ) (optimize using PRO) 11: Output: Pareto-optimal weights w p {p s 1 , . . . , p sn } which are points on the Pareto frontier. The user then chooses one solution by making a trade-off between the performance gains across different metrics. However, as noted earlier this a posteriori choice ignores other solutions that are indistinguishable from the chosen one. We alleviate this by complementing PMO with ensemble decoding, which we call PMO ensemble, in which each point in the Pareto solution is a distinct component in the ensemble decoder. This idea can also be used in other MMO approaches such as linear combination of metrics (g wavg (.)) mentioned above. In this view, PMO ensemble is a special case of ensemble combination, where the decoding is performed by an ensemble of optimal solutions. The ensemble combination model introduces new hyperparameters β that are the weights of the ensemble components (meta weights). These ensemble weights could set to be uniform in a naïve implementation. Or the user can encode her beliefs or expectations about the individual solutions {p s 1 , . . . , p sn } to set the ensemble weights (based on the relative importance of the components). Finally, one could also include a meta-level tuning step to set the weights β. The PMO ensemble approach is graphically illustrated in Figure 2 The illustration is based on two metrics, metric-1 and metric-2, but could be applied to any number of metrics. Without loss of generality we assume accuracy metrics, i.e. higher inal PMO-PRO seeks to maximize the points on the Pareto frontier (blue curve in the figure) leading to Pareto-optimal solutions. On the other hand, the PMO ensemble combines the different Paretooptimal solutions and potentially moving in the direction of dashed (green) arrows to some point that has higher score in either or both dimensions. Lateen MMO Lateen EM has been proposed as a way of jointly optimizing multiple objectives in the context of dependency parsing (Spitkovsky et al., 2011). It uses a secondary hard EM objective to move away, when the primary soft EM objective gets stuck in a local optima. The course correction could be performed under different conditions leading to variations that are based on when and how often to shift from one objective function to another during optimization. The lateen technique can be applied to the multimetric optimization in SMT by treating the different metrics as different objective functions. While the several lateen variants are also applicable for our task, our objective here is to improve performance across the different metrics (being optimized). Thus, we restrict ourselves to the style where the search alternates between the metrics (in round-robin fashion) at each iteration. Since the notion of convergence is unclear in lateen setting, we stop after a fixed number of iterations optimizing the tuning set. In terms of Figure 1, lateen MMO corresponds to alternately maximizing the metrics along two dimensions as depicted by the solid arrows. By the very nature of lateen-alternation, the metric score is better. weights obtained at each iteration are likely to be best for the metric that was optimized in that iteration. Thus, one could use weights from the last k iterations (for lateen-tuning with as many metrics) and then decode the test set with an ensemble of these weights as in PMO ensemble. However in practice we find the weights to converge and we simply use the weights from the final iteration to decode the test set in our lateen experiments. Union of Metrics At each iteration lateen MMO excludes all but one metric for optimization. An alternative would be to consider all the metrics at each iteration so that the optimizer could try to optimize them jointly. This has been the general motivation for considering the linear combination of metrics (Cer et al., 2010;Servan and Schwenk, 2011) resulting in a joint metric, which is then optimized. However due to the scaling differences between the scores of different metrics, the linear combination might completely suppress the metric having scores in the lower-range. As an example, the RIBES scores that are typically in the high 0.7-0.8 range, dominate the BLEU scores that is typically around 0.3. While the weighted linear combination tries to address this imbalance, they introduce additional parameters that are manually fixed and not separately tuned. We This is similar to the PMO-PRO approach except that here the optimizer tries to simultaneously maximize the number of high scoring points across all metrics. Thus, instead of the entire Pareto frontier curve in Figure 1, the union approach optimizes the two dimensions simultaneously in each iteration. Ensemble Tuning These methods, even though novel, under utilize the power of ensembles as they combine the solution only at the end of the tuning process. We would prefer to tightly integrate the idea of ensembles into the tuning. We thus extend the ensemble decoding to ensemble tuning. The feature weights are replicated separately for each evaluation metric, which are treated as components in the ensemble decoding and tuned independently in the optimization step. Initially the ensemble decoder decodes a devset using a weighted ensemble to produce a single N-best list. For the optimization, we employ a two-step approach of optimizing the feature weights (of each ensemble component) followed by a step for tuning the meta (component) weights. The optimized weights are then used for decoding the devset in the next iteration and the process is repeated for a fixed number of iterations. Modifying the MMO representation in Equation 3, we formulate ensemble tuning as: Here the ensemble decoder function N ens (.) is parameterized by an ensemble of weights w M 1 , . . . , w M k (denoted as {w M } in Eq 5) for each metric and a mixture operation (⊗). λ represents the weights of the ensemble components. Pseudo-code for ensemble tuning is shown in Algorithm 2. In the beginning of each iteration (line 2), the tuning process ensemble decodes (line 4) the tuning set using the weights obtained from the previous iteration. Equation 5 gives the detailed expression for the ensemble decoding, where H ens denotes the N-best list generated by the ensemble decoder. The method now uses a dual tuning strategy involving two phases to optimize the weights. In the first step it optimizes each of the k metrics independently (lines 6-7) along its respective dimension in λ ← PMO-PRO(H ens , {w M }) (Alg 1) 10: Output: Optimal weights {w M } and λ the multi-metric space (as shown by the solid arrows along the two axes in Figure 1). This yields a new set of weights w * for the features in each metric. The second tuning step (line 9) then optimizes the meta weights (λ) so as to maximize the multimetric objective along the joint k-dimensional space as shown in Equation 7. This is illustrated by the dashed arrows in the Figure 1. While g(.) could be any function that combines multiple metrics, we use the PMO-PRO algorithm (Alg. 1) for this step. The main difference between ensemble tuning and PMO ensemble is that the former is an ensemble model over metrics and the latter is an ensemble model over Pareto solutions. Additionally, PMO ensemble uses the notion of ensembles only for the final decoding after tuning has completed. Implementation Notes All the proposed methods fit naturally within the usual SMT tuning framework. However, some changes are required in the decoder to support ensemble decoding and in the tuning scripts for optimizing with multiple metrics. For ensemble decoding, the decoder should be able to use multiple weight vectors and dynamically combine them according to some desired mixture operation. Note that, unlike , our approach uses just one model but has different weight vectors for each metric and the required decoder modifications are simpler than full ensemble decoding. While any of the mixture operations proposed by could be used, in this pa-per we use log-wsum -the linear combination of the ensemble components and log-wmax -the combination that prefers the locally best component. These are simpler to implement and also performed competitively in their domain adaptation experiments. Unless explicitly noted otherwise, the results presented in Section 6 are based on linear mixture operation log-wsum, which empirically performed better than the log-wmax for ensemble tuning. Experiments We evaluate the different methods on Arabic-English translation in single as well as multiple references scenario. Corpus statistics are shown in Table 1. For all the experiments in this paper, we use Kriya, our in-house Hierarchical phrasebased (Chiang, 2007) (Hiero) system, and integrated the required changes for ensemble decoding. Kriya performs comparably to the state of the art in phrase-based and hierarchical phrase-based translation over a wide variety of language pairs and data sets . We use PRO (Hopkins and May, 2011) for optimizing the feature weights and PMO-PRO (Duh et al., 2012) for optimizing meta weights, wherever applicable. In both cases, we use SVM-Rank (Joachims, 2006) as the optimizer. We used the default parameter settings for different MT tuning metrics. For METEOR, we tried both METEOR-tune and METEOR-hter settings and found the latter to perform better in BLEU and TER scores, even though the former was marginally better in METEOR 3 and RIBES scores. We observed the margin of loss in BLEU and TER to outweigh the gains in METEOR and RIBES and we chose METEOR-hter setting for both optimization and evaluation of all our experiments. Evaluation on Tuning Set Unlike conventional tuning methods, PMO (Duh et al., 2012) was originally evaluated on the tuning set to avoid potential mismatch with the test set. In order to ensure robustness of evaluation, they redecode the devset using the optimal weights from the last tuning iteration and report the scores on 1-best candidates. Corpus Training size Tuning/ test set We follow the same strategy and compare our PMO-ensemble approach with PMO-PRO (denoted P) and a linear combination 4 (denoted L) baseline. Similar to Duh et al. (2012), we use five different BLEU:RIBES weight settings, viz. Figure 2(a) shows the Pareto frontier of L and P baselines using BLEU and RIBES as two metrics. The frontier of the P dominates that of L for most part showing that the PMO approach benefits from picking Pareto points during the optimization. We use the PMO-ensemble approach to combine the optimized weights from the 5 tuning runs and re-decode the devset employing ensemble decoding. This yields the points LEns and PEns in the plot, which obtain better scores than most of the individual runs of L and P. This ensemble approach of combining the final weights also generalizes to the unseen test set as we show later. Figure 2(b) plots the change in BLEU during tuning in the multiple references and the single reference scenarios. We show for each baseline method L and P, plots for two different weight settings that obtain high BLEU and RIBES scores. In both datasets, our ensemble tuning approach dominates the curves of the (L and P) baselines. In summary, these results confirm that the ensemble approach achieves results that are competitive with previous MMO methods on the devset Pareto curve. We now provide a more comprehensive evaluation on the test set. Evaluation on Test Set This section contains multi-metric optimization results on the unseen test sets, one test set has multiple references and the other has a single-reference. We plot BLEU scores against other metrics (RIBES, METEOR and TER) and this allows us to compare the performance of each metric relative to the defacto standard BLEU metric. Baseline points are identified by single letters B for BLEU, T for TER, etc. and the baseline (singlemetric optimized) score for each metric is indicated by a dashed line on the corresponding axis. MMO points use a series of single letters referring to the metrics used, e.g. BT for BLEU-TER. The union of metrics method is identified with the suffix 'J' and lateen method with suffix 'L' (thus BT-L refers to the lateen tuning with BLEU-TER). MMO points without any suffix use the ensemble tuning approach. Figures 3 and 4(a) plot the scores for the MTA test set with 4-references. We see noticeable and some statistically significant improvements in BLEU and RIBES (see Table 2 for BLEU improvements). All our MMO approaches, except for the union method, show gains on both BLEU and RIBES axes. Figures 3(b) and 4(a) show that none of the proposed methods managed to improve the baseline scores for METEOR and TER. However, several of our ensemble tuning combinations work well for both ME-TEOR (BR, BMRTB3, etc.) and TER (BMRT and BRT) in that they improved or were close to the baseline scores in either dimension. We again see in these figures that the MMO approaches can improve the BLEU-only tuning by 0.3 BLEU points, without much drop in other metrics. This is in tune with the finding that BLEU could be tuned easily (Callison-Burch et al., 2011) and also explains why it remains a popular choice for optimizing SMT systems. Among the different MMO methods the ensemble tuning performs better than lateen or union approaches. In terms of the number of metrics being optimized jointly, we see substantial gains when using a small number (typically 2 or 3) of metrics. Results seem to suffer beyond this number; probably because there might not be a space that contain solution(s) optimal for all the metrics that are jointly optimized. We hypothesize that each metric correlates well (in a looser sense) with few others, but not all. For example, union optimizations BR-J and BMT-J perform close to or better than RIBES and TER baselines, but get very poor score in METEOR. On the other hand BM-J is close to the METEOR baseline, while doing poorly on the RIBES and TER. This behaviour is also evident from the single-metric baselines, where R and T-only settings are clearly distinguished from the M-only system. It is not clear if such distinct classes of metrics could be bridged by some optimal solution and the metric dichotomy requires further study as this is key to practical multimetric tuning in SMT. The lateen and union approaches appear to be very sensitive to the number of metrics and they generally perform well for two metrics case and show degradation for more metrics. Unlike other approaches, the union approach failed to improve over the baseline BLEU and this could be attributed to the conflict of interest among the metrics, while choosing example points for the optimization step. The positive example preferred by a particular metric could be a negative example for the other metric. This would only confuse the optimizer resulting in poor solutions. Our future line of work would be to study the effect of avoiding such of conflicting examples in the union approach. For the single-reference (ISI) dataset, we only plot the BLEU-TER case in Figure 4(b) due to lack of space. The results are similar to the multiple references set indicating that MMO approaches are equally effective for single references 5 . shows the BLEU scores for our ensemble tuning method (for various combinations) and we again see improvements over the baseline BLEU-only tuning. Human Evaluation So far we have shown that multi-metric optimization can improve over single-metric tuning on a single metric like BLEU and we have shown that our methods find a tuned model that performs well with respect to multiple metrics. Is the output that scores higher on multiple metrics actually a better translation? To verify this, we conducted a post-editing human evaluation experiment. We compared our ensemble tuning approach involving BLEU, METEOR and RIBES (B-M-R) with systems optimized for BLEU (B-only) and METEOR (M-only). We selected 100 random sentences (that are at least 15 words long) from the Arabic-English MTA (4 references) test set and translated them using the three systems (two single metric systems and BMR ensemble tuning). We shuffled the resulting translations and split them into 3 sets such that each set has equal number of the translations from three systems. The translations were edited by three human annotators in a post-editing setup, where the goal was to edit the translations to make them as close to the references as possible, using the Post-Editing Tool: PET (Aziz et al., 2012). The annotators were not Arabic-literate and relied only on the reference translations during post-editing. The identifiers that link each translation to the system that generated it are removed to avoid annotator bias. In the end we collated post-edited translations for each system and then computed the system-level erence sentence. Our experiment shows that even with a single reference MMO methods can work. human-targeted (HBLEU, HMETEOR, HTER) scores, by using respective post-edited translations as the reference. First comparing the HTER (Snover et al., 2006) scores shown in Table 3, we see that the single-metric system optimized for ME-TEOR performs slightly worse than the one optimized for BLEU, despite using METEOR-hter version (Denkowski and Lavie, 2011). Ensemble tuning-based system optimized for three metrics (B-M-R) improves HTER by 4% and 6.3% over BLEU and METEOR optimized systems respectively. The single-metric system tuned with M-only setting scores high on HBLEU, closely followed by the ensemble system. We believe this to be caused by chance rather than any systematic gains by the Monly tuning; the ensemble system scores high on HMETEOR compared to the M-only system. While HTER captures the edit distance to the targeted reference, HMETEOR and HBLEU metrics capture missing content words or synonyms by exploiting n-grams and paraphrase matching. We also computed the regular variants (BLEU, METEOR and TER), which are scored against original references. The ensemble system outperformed the single-metric systems in all the three metrics. The improvements were also statistically significant at p-value of 0.05 for BLEU and TER.	2014-07-01T00:00:00.000Z	2013-06-01T00:00:00.000	{ "year": 2013, "sha1": "64c353ac75b09027d4e4fda568e7811719be1876", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "ACL", "pdf_hash": "8afd759ee11f742fbc380badd3c29a89969782ba", "s2fieldsofstudy": [ "Computer Science", "Linguistics" ], "extfieldsofstudy": [ "Computer Science" ] }
269785919	pes2o/s2orc	v3-fos-license	Analysis of Marketing Strategies in Increasing Patient Satisfaction Using SWOT Analysis at Voluntary Healthy Clinic in Pondok Aren, South Tangerang This research aims to find out what kind of marketing strategy should be used to increase patient satisfaction and also to identify the strengths, weaknesses, opportunities and threats that exist at the Voluntary Healthy Clinic. This research uses qualitative research with a case study approach. Primary data was collected through interviews with several clinic stakeholders. Apart from using the interview method, the author also uses observation and documentation methods to deepen the data obtained. The data that has been collected is then analyzed using the SWOT method. The results of this research show that an effective marketing strategy used to increase patient satisfaction is at quadrant 1 point, namely supporting aggressive strategies (SO/Strengths and Opportunities) by implementing several strategies, namely (1) By utilizing a strategic location, the clinic can take advantage of various opportunities such as develop services or programs, namely laboratories, midwifery, inpatient services, etc. and carry out promotions via internet media, distribute brochures and put up banners, (2) With strengths such as ambulance facilities, various kinds of patent medicines, home services care/home visits, long-distance consultations provide opportunities to increase patient satisfaction and also increase patient confidence in seeking treatment at the Voluntary Healthy Clinic. Research Background In recent years, many people have paid attention to the Health sector.One of the main concerns is special examination services such as swab tests which are sought after by many consumers.However, it is still difficult to find a swab test service provider that is easy to reach at an affordable price. A clinic is a medical facility that has a smaller scope and focuses on certain complaints and health services for patients who do not require hospitalization.Its presence in the community has important value because clinics provide health services that are almost equivalent to hospitals, but at more affordable costs. In this research, the clinic in question is the Voluntary Healthy Clinic (KSS) which is supported by the Taman Mandiri Syariah Foundation (YTS), which is an organization or business entity that operates in the health services sector.Voluntary Healthy Clinic (KSS) prioritizes patient or service user satisfaction by providing quality services and supported by competent and professional staff.Apart from that, KSS also pays attention to patient welfare.Optimal service will not be achieved without a sense of shared ownership and responsibility. Formulation of the problem Based on the background description above, the research focus in this study includes: 1. What marketing strategy should be implemented to increase patient satisfaction at the Voluntary Healthy Clinic? 2. How to identify the strengths, weaknesses, opportunities and threats that exist at the Voluntary Healthy Clinic? Management Understanding Management Definition of management according to Ramdan, T,. & Sufyani, MA (2019:20) Management is a science that studies the management of organizational resources effectively and efficiently within the framework of objectives through the processes of planning, organizing, directing and supervising. Marketing Management Understanding Marketing Management "(Marketing Management is the planning, direction and control of the entire marketing activity of a firm or division of a firm) marketing management is planning, directing and supervising all marketing activities (Shultz in Manap, 2016:79)" Marketing Strategy Marketing Strategy Definition In marketing a product, a strategy is needed so that we can market the product effectively and efficiently so that the targets and objectives of the marketing itself can be achieved.The definition of marketing strategy according to Fandy Tjiptono (2017:228) "marketing strategy is a plan that outlines the company's expectations of the impact of various 4 marketing activities or programs on demand for its product or product line in a particular target market" Segmenting, Targeting and Positioning The STP (Segmentation, Targeting, Positioning) strategy is a company's initial plan to dive into marketing.The main objective of segmentation, targeting and positioning strategies is to position a brand in the minds of consumers in such a way that the brand has a sustainable competitive advantage.a. Understanding Segmenting According to Kotler & Armstrong (in Perdianto Jon, 2017) say that through market segmentation, companies will divide large and heterogeneous markets into smaller segments that can be served efficiently with products and services to meet them.According to Indriyo Gitosudarmo (2014); Market segmentation is an attempt to group markets, from markets that are heterogeneous to parts of the market that are homogeneous. Based on the definition above, it can be concluded that market segmentation is the process of dividing the market into certain groups or segments according to different characteristics, needs and behavior. 1). Basics of Segmentation The basics of segmenting depend greatly on the market to which the segment is applied.Marketers must be able to differentiate one market segment from another market segment.Segmenting can be achieved using several different methods.This method can also differ from one product to another. 2). How to Do Segmenting According to Suprayanto and Rosad (in Sudartono, 2019) there are several criteria that must be met by market segments so that the market segmentation process can be carried out effectively and be beneficial for the company, namely: a) Different (distinctive) shows that the segment has characteristics and purchasing attitudes that are different from other segments.b) Measurability (measurability) shows that the purchasing power of each segment must be able to be measured to a certain level even though in reality it is a certain variable that is not easy to measure.c) Achievable (accessibility) shows how far the segment can be reached and served efficiently.d) Meaning (substantiality) that a group will be worthy of being called a segment if it is large enough and can be profitable.e) Feasibility shows how far efficient programs can be prepared and attract segment attention.f) Can provide benefits (profitability) for a target market segment that can bring financial benefits to the industry, either directly or indirectly. 3). Purpose of Segmenting The purpose of segmentation is to understand more effectively and efficiently the needs, characteristics or behavior of different consumers who may require a separate product or marketing mix.According to Kasali (in Mujahidin, Khoirianingrum 2019) there are 5 benefits obtained by segmenting the market, including: a) Designing products that are more responsive to market needs.b) Analyze the market c) Find opportunities d) Mastering a superior and competitive position e) Determine effective and efficient communication strategies Patient satisfaction Understanding Patient satisfaction According to Pohan, ( 2018) who states patient satisfaction if the performance of the health services they receive is the same or exceeds their expectations and vice versa.When patients feel satisfied, it can help speed up the service process provided. Understanding SWOT Analysys SWOT analysis is the systematic identification of various factors to formulate company strategy, especially in terms of company marketing strategy.Marketing strategy is a series of stimuli placed in the consumer's environment that are designed to influence consumer affection, cognition and behavior (Peter and Olson, 2014). According to Erwin Suryatama (2016: 130) says that "SWOT analysis is a strategic planning method used to evaluate strengths, weaknesses and weaknesses, opportunities or opportunities, and threats in a project or business speculation and can be applied by analyzing and sorting out various things that influence the four factors" Types of research The research that the researcher conducted used qualitative research with a case study approach.According to Sugiyono (2019: 7-8), the qualitative research method is a new method, because of its recent popularity, it is called the postpositivistic method because it is based on postpositivistic philosophy.This method is also called an artistic method, because the research process is more artistic (less patterned), and is called an interpretive method because research data is more concerned with the interpretation of data determined in the field.Sugiyono (2019:8) said that qualitative research methods are often called naturalistic research methods because the research is carried out in natural conditions (natural settings); also called the ethnographic method, because initially this method was more widely used for research in the field of cultural anthropology; It is called a qualitative method, because the data collected and the analysis is more qualitative in nature. The place of research on Marketing Strategy Analysis in Increasing Patient Satisfaction Using SWOT Analysis was carried out at the Voluntary Healthy Clinic in Pondok Aren which is located at JL. Raya Pondok Aren, Ruko Arinda Permai I Block A1 No.9 Pondok Aren Village, Pondok Aren District, South Tangerang Instrument Research One of the characteristics of qualitative research is that the researcher acts as both an instrument and a data collector.Non-human instruments (such as: questionnaires, interview guidelines, observation guidelines, documentation guidelines and so on) can also be used, but their function is limited to supporting the researcher's task as a key instrument.Therefore, in qualitative research, the presence of the researcher is absolute, because the researcher must interact with both human and non-human environments in the research arena.His presence in the research field must be explained whether his presence is known or unknown to the research subject.This relates to the researcher's involvement in the research arena, whether actively or passively involved.(Pure, 2017). In qualitative research, the main instrument in data collection is humans, namely, the researcher himself or other people who help the researcher.In qualitative research, the researcher himself collects data by asking, requesting, listening, and taking.Researchers can ask for help from other people to collect data, called interviewers.In this research, researchers used interview, observation and documentation research instruments. Interview Interview or debriefing is a research technique carried out by means of dialogue either directly (face to face) or remotely via other communication media channels (Sanjaya, 2015:263) Observation Observation in research is defined as focusing attention on an object by involving all the senses to obtain data.Observation is direct observation using sight, smell, hearing, touch, or if necessary, taste.Instruments used in observation can be observation guides, tests, questionnaires, image recordings and sound recordings. Observation instruments are used in qualitative research as a complement to the interview techniques that have been carried out.Observation in qualitative research is used to directly see and observe the research object, so that the researcher is able to record and collect the data needed to reveal the research being conducted.Observations in qualitative research, researchers must first understand the variations in observations and the roles carried out by researchers (Ulfatin, 2014). RESEARCH RESULT EFAS and IFAS In this case, Klinik Sehat Sukarela should implement a Strengths-Opportunities (SO) strategy, where the author combines various strengths with the opportunities available to the clinic Below, I present the SO strategy along with detailed explanations for each point within the strategy: 1. To add patented medicine products so that the benefits of the medicine produced are suitable for patients, in order to build patient trust in the clinic The author combines the clinic's strength, which is the predominant use of patented medicine products, with the clinic's opportunity, which is building trust with potential or current patients The author develops this strategy because although the clinic primarily offers patented medicine products, there are also some generic medicine products available, in order to gain patient trust.The clinic should increase its stock of patented medicines so that patients or potential patients feel comfortable seeking treatment at the clinic In this regard, the author believes that if the clinic expands its range of patented drugs, it will have a better chance of gaining the trust of patients or potential patients, as patented drugs are often more effective than generic ones This could attract patients to seek treatment at Klinik Sehat Sukarela and create an opportunity for Klinik.2. Utilizing a strategic location to develop services or programs such as laboratories, midwifery, and inpatient care.This strategy is the result of a combination of strengths, namely strategic placement location, close to the highway and residential areas, and opportunities, namely developing programs that have not been implemented yet, such as inpatient services, laboratory, midwifery, radiology, pharmacy, etc The author formulated this strategy by utilizing the strategic location to develop services or programs such as laboratory, midwifery, inpatient care, etc, because of the placement location of this clinic being close to road access and residential areas will certainly facilitate the development of programs or services that have not yet been reached and take advantage of the Clinic's location opportunities.3.By utilizing its strategic location, the clinic promotes itself by assessing the opportune moments to spread information through WhatsApp or Instagram, as well as offline methods such as distributing brochures and putting up banners.The author in this strategy combines the Clinic's strengths, namely "Strategic location, close to main roads and residential areas" with opportunities, namely "Spreading promotions with brochures, posters, placing banners and also at certain events or occasions."The author created this strategy because this strategic location can easily carry out offline promotions such as distributing brochures, putting up banners, or when holding certain events such as procuring vaccines, social services or mass circumcisions, there we can also promote services at the same time.at the Clinic.This strategy is a form of promotion that can be carried out by the Clinic to attract patients to seek treatment at the Clinic.4. Using ambulance facilities as much as possible as a means of promotion to increase patient satisfaction and patient trust.This strategy combines the clinic's strengths, namely "Ambulance facilities are available" and opportunities, namely "Building the trust of prospective patients or patients".The author created this strategy because ambulance facilities are available at the Clinic, which can be used by any local resident with permission from the Clinic or the Foundation.This is one means of building the trust of prospective patients and increasing the satisfaction of patients who use the services or services of the Clinic. 5. Promoting home care services which are usually still rare and perhaps not yet programmed by similar clinics to gain patient trust in the clinic.This strategy combines the Clinic's strengths, namely "Services for visiting patients at home or what can be called (Home visit/ Home care)" with the Clinic's opportunities, namely "Building the trust of prospective patients or patients".Currently, clinics have implemented this strategy because they see the strength of clinics that have home care/home visit services (visit patients' homes) where services like this are still rare or perhaps no one provides them like similar clinics, and this is a means to increase satisfaction.patients, increasing patient trust and also improving service to patients and prospective patients.6. Utilize facilities such as being able to consult first by telephone before seeking treatment or procedures at the clinic to gain patient trust and patient satisfaction.The author in this strategy combines the strength of the Clinic, namely "Can carry out consultations first by telephone" with the opportunity of the Clinic, namely "Building the trust of potential patients or patients". Currently the Clinic has implemented this strategy.This strategy is a plus point for the clinic for patients because facilities such as being able to consult first by telephone before seeking treatment or this procedure are rarely found in similar clinics.In this case, the author provides strategies for clinics to improve remote consultation service facilities to gain the trust and satisfaction of patients and prospective patients. CONCLUSIONS AND RECOMMENDATIONS From the results of the analysis, description and research at the Pondok Aren Voluntary Health Clinic on the external and internal strategic environment, several conclusions can be put forward as follows: 1.Based on the internal and external analysis results presented in the form of IFAS and EFAS matrices, with a total weighted IFAS score of 319 and an EFAS score of 319, the current position of the company is in quadrant I This condition indicates that the clinic is in a very favorable situation The company has many strengths and opportunities Therefore, Klinik Sehat Sukarela should utilize its strengths to capitalize on opportunities and enhance patient satisfaction The appropriate strategy to be implemented by Klinik Sehat Sukarela to improve patient satisfaction is an aggressive approach, seizing existing opportunities and competing effectively in the business world, while supporting aggressive growth policies.2. The results of the SWOT analysis for Klinik Sehat Sukarela reveal the following strengths: providing treatment for orphaned and underprivileged patients at voluntary rates, offering telephone consultations, having ambulance facilities available, providing home visits or home care services, using predominantly patented medicines, strategically located near highways and residential areas However, there are weaknesses such as medication and treatment costs not covered by BPJS/insurance, incomplete and inadequate facilities, limited payment methods (only via transfer or cash), reliance on general pharmacies for medication orders, and the absence of a pharmacist As for opportunities.Based on the above conclusions, the suggestions that I can convey and hope will be beneficial and considered by Klinik Sehat Sukarela are as follows: 1. Implement marketing strategies that have been analyzed by the author using SWOT analysis, namely: a. Adding patented pharmaceutical products to ensure that the medications produced are suitable for patients, in order to build patient trust in seeking treatment at the clinic b.Utilizing the strategic location to expand services or programs such as laboratory, midwifery, and inpatient services c.Leveraging the strategic location, the clinic can promote itself by choosing the right timing to spread information through platforms like WhatsApp or Instagram, as well as offline methods such as distributing brochures and displaying banners d.Maximizing the use of ambulance facilities as a promotional tool to enhance patient satisfaction and trust e.Utilizing facilities such as pre-consultation via telephone before receiving treatment or procedures at the clinic can help build patient trust and satisfaction 2. The marketing strategy provided by the author is a short-term marketing strategy that will be relevant for the next 1-3 years, as well as some longterm strategies that will serve indefinitely Klinik Sehat Sukarela needs to reassess and determine a marketing strategy that is relevant to the company's situation in the next 1-3 years in order to survive, enhance its business strength or competitive position, and improve patient satisfaction.	2024-05-16T15:09:12.935Z	2024-05-01T00:00:00.000	{ "year": 2024, "sha1": "86b68e3f801b71cc06e36dd4de261cf2c3a34e6e", "oa_license": "CCBY", "oa_url": "https://journal.formosapublisher.org/index.php/ministal/article/download/9002/8948", "oa_status": "HYBRID", "pdf_src": "Anansi", "pdf_hash": "20c5a4ec7572aee55deabc4653791462a17fb1fe", "s2fieldsofstudy": [ "Business", "Medicine" ], "extfieldsofstudy": [] }
244570739	pes2o/s2orc	v3-fos-license	Infantile status epilepticus disrupts myelin development Temporal lobe epilepsy (TLE) is the most prevalent type of epilepsy in adults; it often starts in infancy or early childhood. Although TLE is primarily considered to be a grey matter pathology, a growing body of evidence links this disease with white matter abnormalities. In this study, we explore the impact of TLE onset and progression in the immature brain on white matter integrity and development utilising the rat model of Li-pilocarpine-induced TLE at the 12th postnatal day (P). Diffusion tensor imaging (DTI) and Black-Gold II histology uncovered disruptions in major white matter tracks (corpus callosum, internal and external capsules, and deep cerebral white matter) spreading through the whole brain at P28. These abnormalities were mostly not present any longer at three months after TLE induction, with only limited abnormalities detectable in the external capsule and deep cerebral white matter. Relaxation Along a Fictitious Field in the rotating frame of rank 4 indicated that white matter changes observed at both timepoints, P28 and P72, are consistent with decreased myelin content. The animals affected by TLE-induced white matter abnormalities exhibited increased functional connectivity between the thalamus and medial prefrontal and somatosensory cortex in adulthood. Furthermore, histological analyses of additional animal groups at P15 and P18 showed only mild changes in white matter integrity, suggesting a gradual age-dependent impact of TLE progression. Taken together, TLE progression in the immature brain distorts white matter development with a peak around postnatal day 28, followed by substantial recovery in adulthood. This developmental delay might give rise to cognitive and behavioural comorbidities typical for early-onset TLE. Supplementary methods On postnatal day (P) 11, the animals were marked for identification and randomly assigned to SE or control groups, ensuring that each group contained animals from multiple litters to exclude a litter effect. Each litter consisted of both controls and SE exposed animals. In the group of additional animals for histology, the rats were at the same time randomly allocated for tissue collection at postnatal days 15, 18, or 28 (P15, P18, and P28) while controlling for the litter effect. All pups received LiCl (127 mg/kg; LiCl 127 mg was dissolved in 5 ml of water for injection and injected intraperitoneally 127 mg/5ml/kg) and were returned to their dams. On P12 pups were separated from their dams and transferred into a special silent room with controlled conditions. Experiments were always performed at the same time of the day, between 10 am and 2 pm (i.e. during the light period). Animals were placed individually into small containers and they were maintained at +33±1°C to compensate for the immature thermoregulation at this age (Conklin and Heggeness, 1971) during the entire period of separation from their mothers. SE was induced with a single intraperitoneal injection of pilocarpine (35mg/kg; 35 mg of pilocarpine was dissolved in 5 ml of saline and injected). Controls received corresponding volume of saline instead. The severity of motor SE was assessed using the following scoring system: 0 -normal behaviour 1 -stereotypic behaviour (face washing, scratching), isolated myoclonic jerks 2 -head bobbing, pivoting, swimming movements 3 -clonic seizures with preserved righting reflexes 4 -repeated periods of wild running 5 -generalized tonic-clonic seizures with loss of righting reflexes. Animals were assigned a score for the most severe behavioural characteristics. Latency to the onset of motor seizures was recorded. Mortality was recorded throughout the entire experimental period. Only rats that exhibited behavioural manifestations of seizures progressing to forelimbs clonus (i.e., score 3) for at least 1 hour and without periods of wild running and generalized tonic-clonic seizures (score 4-5) were used for further experiments. After 1.5 hours of convulsive SE, animals were given a single dose of paraldehyde (0.07ml/kg; 0.07 ml of paraldehyde was mixed with 5 ml of saline and injected intraperitoneally). Controls received paraldehyde in the same dose. Approximately 30 min later they were injected subcutaneously with 0.5 ml saline to restore the volume loss. After the brief recovery, pups were returned to their dams (the duration of isolation from mothers in the control and SE groups was the same ~ 4 hours). The weight of pups was checked daily and animals that did not gain any weight within 24hrs after SE were given 0.5ml of saline with 10% of glucose subcutaneously to prevent dehydration and further weight loss. To minimize the effects of variability in individual groups, the data were used to calculate relative body weight (body weight at P11 was taken as 100%). The difference in the relative body weights between two consecutive days (starting at P11) was used as a measure of weight gain ( Figure S4). Statistical comparison of relative weight gains between controls and SE animals was done using RStudio (version 1.2.5033). Data were compared using Mixed-effects analysis with False Discovery Rate (FDR) correction for multiple comparisons. Mixed effect analysis was used instead of One-way repeated measure ANOVA because of missing values. Statistical disclaimer: Multiple testing correction techniques are known to reduce false-positive discovery while increasing the risk of type II error (false-negative results) (Drachman, 2012;Ranganathan et al., 2016;White et al., 2019). To avoid the possible unnecessary introduction of false-negative values in cases when the likelihood of false-positive discovery is below 1 per compared set, we consider p-value without adjustment (displayed in table S1 and S2) a true p-value in our ROI statistics with less than 20 comparisons per method (S1) or position (S2). Figure S1 -ROIs representing the default mode network. Locations of regions of interest used for the analysis of default mode network connectivity in rat brain fMRI; slice thickness 1 mm. ROIs are overlaid on the reference brain, which was acquired with a functional spin-echo echo planar imaging sequence. Numbers located under individual brains indicate distance from the bregma in mm. Figure S2 -Brain sections for analysis of optical density. Illustrative positions of Black-Gold IIstained brain sections selected for analysis of optical density in large white matter structures with indicated approximate distance from the bregma in the adult animal. Coronal brain sections from 15-, 18-, 28-, and 72-day-old rats are displayed. Figure S3 -Position of regions of interest (ROIs) for the optical density quantification. Representative positions of ROIs in white matter structures in left and right hemisphere Figure S4 -Animal weight. Daily weight percentage change (A) of rats following the induction of status epilepticus (SE) and controls (Ctrl). Daily follow-up is displayed across 16 days starting at the day of SE induction (0). The symbols represent the mean value and the error bars represent the mean standard deviation. The red asterisk (*) indicates significantly different values when comparing SE and control rats using Mixedeffects analysis. Displayed data include weight information from all animals used in the study (both MRI and histology groups). Graph (B) shows absolute animal weight [g] on the day of MRI & fMRI data collection (postnatal day (P) 28, 72 and 75) for animals after SE and controls within the MRI group.	2021-11-25T14:30:26.961Z	2021-11-01T00:00:00.000	{ "year": 2021, "sha1": "07b6fd3e658fb82e94cc227942246e1f4eb27790", "oa_license": "CCBY", "oa_url": "https://doi.org/10.1016/j.nbd.2021.105566", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "30bd2c0471b914fb46af2fc560bbe804cbba3533", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [] }
247079346	pes2o/s2orc	v3-fos-license	Cryptogenic organizing pneumonia associated with pregnancy: A case report BACKGROUND Cryptogenic organizing pneumonia (COP), formerly known as bronchiolitis obliterans organizing pneumonia, is an extremely rare disease in pregnancy. In this case, we report on COP diagnosed in recurrent pneumonia that does not respond to antibiotics in pregnant woman. CASE SUMMARY A 35-year-old woman with no prior lung disease presented with concerns of chest pain with cough, sputum, dyspnea, and mild fever at 11 wk’ gestation. She was diagnosed with community-acquired pneumonia and treated with antibiotics; her symptoms improved temporarily. Four weeks after discharge, she was re-admitted with aggravated symptoms. Chest computed tomography demonstrated multifocal patchy airspace consolidation and ground-glass opacities at the basal segments of the right lower lobe, at the lateral basal segment of the lower lobe, and at the lingular segment of the left upper lobe. Bronchoalveolar lavage revealed an increased lymphocyte count and a decreased CD4/CD8 ratio. Prednisolone (0.5 mg/kg/d) was administered for 10 d after the second admission. Dyspnea improved after 3 d of steroid treatment and other symptoms improved on the 5th day of steroid administration. Post-delivery transbronchial lung biopsy further revealed the presence of granulation tissue with fibroblasts in small-bronchiole lumens. CONCLUSION This case suggests that it is important to differentiate COP from atypical pneumonia in the deteriorated condition despite antibiotic treatment. INTRODUCTION Cryptogenic organizing pneumonia (COP) is a diffuse infiltrating lung disease, wherein granulation tissue proliferates in the small bronchiolar epithelium damaged owing to various causes and consequently obstructs alveolar ducts and alveoli [1,2]. The occurrence of COP in pregnancy is extremely rare and pregnancy-related physiological changes may worsen respiratory complications in COP. Previously reported cases had pre-onset underlying diseases such as asthma, fungal infection and Crohn's disease that can cause inflammatory condition (Table 1). Here, we report the fourth case of COP in a pregnant woman without underlying medical history initially diagnosed with community-acquired pneumonia that did not improve with antibiotic treatment. Chief complaints A 35-year-old woman, gravida 2, para 1, presented with concerns of chest wall pain with cough, sputum, dyspnea, and mild fever of 37.7 °C at 11 wk of gestation. History of present illness The mild cough was started ten days ago with gradually aggravated feature. History of past illness Her obstetric history included spontaneous vaginal delivery at 40 wk of gestation, with no special medical history. In particular, there was no history of previous pulmonary diseases. A 7.5 pack-year history of smoking was noted before the first pregnancy by an antenatal evaluation. Physical examination On admission, the patient's blood pressure was 120/80 mmHg, body temperature was 37.7 °C, pulse rate was 108/min, oxygen saturation was 96%, and respiratory rate was 28/min with a rale in the right lower lung area. Imaging examinations Chest radiography showed increased patchy opacities in the right lower lobe (Figure 1), and computed tomography (CT) revealed some patchy lobular consolidation and peripheral ground-glass opacities (GGOs) in the posterior and lateral basal segments of the right lower lobe ( Figure 2). The pulmonary function test showed a forced vital capacity (FVC) of 2.87 L (77% of predicted), forced expiratory volume in one second (FEV1) of 2.35 L (74% of predicted), FEV/FVC ratio of 82%, and peak expiratory flow of 6.19 L/s. The tidal flow-volume curve revealed minimal obstructive lung disease. An ultrasound examination showed that appropriate fetal growth for the gestational age, a normal amount of amniotic February 26, 2022 Volume 10 Issue 6 FINAL DIAGNOSIS The increased lymphocyte count (40%) and a decrease in the CD4/CD8 ratio (0.6) with the presence of macrophages (25%) and neutrophils (8%) in BAL suggested a diagnosis of COP. TREATMENT We began steroid treatment with prednisolone (0.5 mg/kg/d), and progressive improvement of radiological findings was noted. Dyspnea improved after 3 d of steroid treatment, and other symptoms were reduced on the 5 th day of steroid administration. OUTCOME AND FOLLOW-UP Post-discharge, the patient did not express any special events during pregnancy and gave birth by vaginal delivery at 39+4 d of gestation (male, 3370 g; Apgar scores of 8 and 9 at 1 and 5 min, respectively). Transbronchial lung biopsy was conducted after delivery without any complications, and the proliferation of granulation tissue into the bronchioles and alveolar duct indicated COP (Figure 3). DISCUSSION The prevalence of COP is unknown and is mainly observed in individuals aged 50-60 years [3]. COP occurrence during pregnancy is extremely rare, but it could be more severe owing to physiologic changes in pregnant women, such as an elevated diaphragm, increased oxygen demand, decreased functional residual capacity, and decreased chest wall compliance [4]. Thus, previous reports have recommended close antenatal care and regular pulmonary function tests to reduce respiratory complications during pregnancy; further, elective preterm delivery can be an option in more severe cases [5]. The clinical features of COP in the case described above were not notably different from the general clinical features of COP. The respiratory symptoms began with a flu-like illness with cough, mild fever, malaise and progression of shortened breathing to dyspnea [2]. Although a quarter of patients with COP had no special physical findings [3], inhalation rales or crackling were observed in the physical examination in this case. In general, imaging approaches are employed to diagnose COP. Chest radiography for COP has three characteristic features: multiple alveolar opacities (typical COP), solitary opacity (focal COP), and infiltrative opacities (infiltrative COP). Bilateral multiple opacities are more common than solitary patterns [6,7]. In our patient, bilateral patchy opacities were observed in both lower lobes. Thin-section CT scans have a correct diagnosis rate of 79% with histologically proved COP [8]. CT findings for COP are patchy GGOs in the subpleural and/or peribronchovascular area (80%), airspace consolidation in bilateral lower lobes (71%), wall thickening and cylindrical dilatation of air bronchogram (71%), ill-defined small nodular opacities (50%), and pleural effusion (in a third of patients)[9]. The specific multifocal patchy February 26, 2022 Volume 10 Issue 6 airspace consolidation, GGOs, and bilateral pleural effusion were observed. Corticosteroids are administered as the initial treatment for COP and are effective for both typical and focal COP. The recommended treatment regimens include initial dosages of 0.75-1.5 mg/kg prednisolone for 3 mo with gradual reduction according to clinical symptom improvement [10]. In this case, we began with a low initial oral dose of prednisolone of 0.5 mg/kg/d after the patient's second admission because corticosteroid use in first trimester can be associated with the development of an orofacial cleft. Fortunately, symptoms improved after 5 d of the low-dose administration and maintenance therapy was continued for 5 more days. This rare case is about the COP diagnosed in pregnant women without underlying medical conditions. In addition, it suggests a diagnostic value of COP, which is less effective in conevntional initial treatment. In this case, a pregnant woman was initially diagnosed with community-acquired pneumonia and treated with antibiotics; her symptoms seemed to improve temporarily but then recurred with greater severity. CONCLUSION COP has similar clinical features with other types of pneumonia and in particular, chest radiographic differentiation of COP could be difficult. The progressive condition indicates a specific clinical aspect of COP; thus, it is important to differentiate COP from other atypical pneumonia that recur despite initial antibiotic treatment.	2022-02-24T16:16:43.560Z	2022-02-26T00:00:00.000	{ "year": 2022, "sha1": "cb27c6861071ee6f19e59e883c35c9754ca966cd", "oa_license": "CCBYNC", "oa_url": "https://doi.org/10.12998/wjcc.v10.i6.1946", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "d202879aa0dc500fa067b215dc874d706b041739", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
93003852	pes2o/s2orc	v3-fos-license	Art and Psychological Well-Being: Linking the Brain to the Aesthetic Emotion Empirical studies suggest that art improves health and well-being among individuals. However, how aesthetic appreciation affects our cognitive and emotional states to promote physical and psychological well-being is still unclear. In this review, we consider the idea that the positive emotional output elicited from the aesthetic experience affects mood, and indirectly promotes health and well-being. First, we examine evidence that arts promoting well-being involve art museums, healthcare settings, and education. Second, we review some neuroimaging studies addressing aesthetic experience and emotional processing. In particular, we leveraged advances in neuroaesthetics to explore different hypotheses about the determinants of aesthetic pleasure during art reception, in the attempt to clarify how experiencing art promotes well-being. Finally, we propose research on aesthetic experience and psychophysiological measures of stress, with the goal of promoting a focused use of art as a tool for improving well-being and health. INTRODUCTION Aesthetic experience concerns the appreciation of aesthetic objects and the resulting pleasure. Such pleasure is not derived from the utilitarian properties of the objects but linked to the intrinsic qualities of the aesthetic objects themselves. Hence, the aesthetic pleasure is disinterested (Kant, 1790). Aesthetic experiences can arise from the appreciation of human artifacts, such as artworks (e.g., poetry, sculpture, music, visual arts, etc.) or aesthetic natural objects like sunsets or mountain vista. In this review, we refer to aesthetic experiences associated with the appreciation of artworks, particularly visual arts. Aesthetic experiences are offered by multiple contexts, (e.g., museums, galleries, churches, etc.). Several psychological perspectives considered aesthetic experience as a rewarding process and suggested a link between aesthetic experience and pleasure (Berlyne, 1974;Leder et al., 2004;Silvia, 2005). Recent studies suggest the arts can promote health and psychological wellbeing and offer a therapeutic tool for many, e.g., adolescents, elderly, and vulnerable individuals (Daykin et al., 2008;Todd et al., 2017;Thomson et al., 2018). Aesthetic experience has been associated with mindfulness meditation, as it leads to enhancing the capability of perceptually engaging with an object (Harrison and Clark, 2016). However, how aesthetic experience affects cognitive and emotional states and promotes physical and psychological well-being is a matter of debate (Daykin et al., 2008). Several theoretical models have been proposed, suggesting alternating key roles for cognitive or emotional facets of the aesthetic experience. A common theme in the models is that the aesthetic evaluation of an artwork is the result of bottom-up stimulus properties and top-down cognitive appraisals (Leder et al., 2004;Chatterjee and Vartanian, 2016;Pelowski et al., 2017). The result affects mood, therefore promoting health and well-being (Kubovy, 1999;Sachs et al., 2015). In this vein, neuroimaging studies highlighted that immediate emotional responses to artwork and low-intensity enduring changes in affective states (cf. Scherer, 2005, for the distinction of emotional response and affective state) are associated with recruitment of brain circuitry involved in emotion regulation, pleasure, and reward. Thus, for instance, images rated as beautiful elicit activity in reward-related areas, such as the medial orbitofrontal cortex, and are associated with higher reward value than those rated as ugly (Kawabata and Zeki, 2004). Moreover, the activation of an emotion processing network comprising the ventral and the dorsal striatum, the anterior cingulate and medial temporal areas has been associated with the transient mood changes in response to happy and sad classical music (Mitterschiffthaler et al., 2007). Here, we review evidence showing that arts promote well-being across several domains, and discuss the neural underpinnings of aesthetic experience, emotional processing, pleasure, and reward. In particular, we assess the idea that a common physiological mechanism underlies aesthetic processing in multiple places for experiencing art. Implications for therapeutic and educational uses of art are discussed. Aesthetic Appreciation and Well-Being Benefits associated with aesthetic processing have been demonstrated in different settings, ranging from reproductions of paintings shown in laboratories to real art contexts such as museums. In the following sections, we present a review of the main research branches on art in which a beneficial effect on health has been shown. Art in the Museum Several studies show benefits of art museums as settings for therapy (Treadon et al., 2006;Chatterjee and Noble, 2013). These benefits include improvement of memory and lower stress levels, and amelioration of social inclusion. Populations studied include older individuals (Salom, 2011;Thomson et al., 2018), people with enduring mental health problems (Colbert et al., 2013), people with dementia (Morse and Chatterjee, 2018), and the socially isolated (Todd et al., 2017). Moreover, in a study with people with dementia and their caregivers, viewing traditional and contemporary galleries, both art sites promoted well-being, including positive social impact and cognitive enhancement (Camic et al., 2014). Research has been conducted to identify the elements of the museum setting that facilitate the treatment goals, including psychological, social, environmental aspects (Salom, 2011;Camic and Chatterjee, 2013;Colbert et al., 2013;Morse and Chatterjee, 2018). Museum environment and artifacts offer an extraordinary aesthetic experience that allows the recollection of positive memories (Biasi and Carrus, 2016), and evidence suggests that these reminiscence activities can affect mood, self-worth, and a general sense of well-being in the elderly (Chiang et al., 2009;O'Rourke et al., 2011;Eekelaar et al., 2012). Museum and galleries, unlike hospitals and clinics, are non-stigmatizing settings. The art setting encourages self-reflection and group communication, facilitating the therapeutic process and thus making them ideal locations for health interventions (Camic and Chatterjee, 2013). Using psychophysiological measures, studies find visits to art museums decrease stress, which could promote health and well-being (Clow and Fredhoi, 2006;Mastandrea et al., 2018). Clow and Fredhoi reported that levels of salivary cortisol and self-reported measure of stress in 28 healthy young individuals decreased significantly after a visit to the Guildhall Art Gallery of London (Clow and Fredhoi, 2006). Similarly, exposure to figurative art lowers systolic blood pressure (SBP), which could have relaxing effects (Mastandrea et al., 2018). Specifically, 64 healthy female participants were assigned to one of three different visits to the National Gallery of Modern Art in Rome: figurative art, modern art, and a control condition consisting of a visit to the museum office. Pre-and post-visit measures of blood pressure and heart rate were acquired, as indices of emotional states associated with the three visit conditions. Results revealed that only figurative art exposure decreased systolic blood pressure. Of interest, participants liked the two art styles equally well, and reduction in SBP was not correlated with liking. In fluency theory, processing ease increases positive emotional response to artwork (Reber et al., 2004). Accordingly, it may be thought that the reduction of levels of ambiguity that characterizes unambiguous figurative arts may have a relaxing effect on the physiological states. On the other hand, as participants in this study were not asked to judge the comprehensibility or hedonic values of artworks, it is not possible to draw firm conclusions about the restorative effects following exposure to figurative, but not abstract artworks in art museum. Art and Education Several studies have been conducted on the efficacy of art-based interventions in professional education, demonstrating a growing interest for this field, and posing challenging opportunities for the traditional learning methods that shape the current teaching practice (Richard, 2007;Leonard et al., 2018). Art-based pedagogy is focused on the integration of an art form (e.g., theatre, visual art-painting, music, etc.) with another subject matter, to enhance learning processes (Rieger and Chernomas, 2013). In learning through art, the learner approaches a subject matter by creating art, responding to art, or performing artistic works not by studying art as a theoretical discipline (Rieger and Chernomas, 2013). This art-based learning (ABL) has been used successfully in healthcare education (Wikström, 2003;Rieger et al., 2016). For instance, using a work of art as a teaching method is effective in increasing students' observational skills, empathy (i.e., abilities in empathizing with the patient and develop compassion), nonverbal communication, and interpersonal relationships, in comparison with traditional teaching programs (Wikström, 2011). Wikstrom (2000) and colleagues showed that Frontiers in Psychology \| www.frontiersin.org an educational program based on visual art dialogue evoked emotional experiences increasing nurses' empathy (Wikstrom, 2000). The students were asked to describe nursing care patterns in the painting "The Sickbed" from Lena Croqvist, after which they were asked strategic questions aimed at eliciting empathetic responses, such as "From a nursing care perspective, how do the characters feel?" A control group was asked to describe good nursing practice without the support of visual art or pictures. The visual art was more effective than the control for expressing aspects of nursing care and in increasing empathy scores (Wikström, 2001). These studies suggest that embedding visual art in healthcare education may increase understanding of emotional experience of chronic pain and suffering of the patients, thereby improving nursing care practices. A limitation of these studies is that control groups received only verbal instruction, that make it difficult to evaluate the specific contribution of art-specific visual support (i.e., visual portraits, artworks, etc.) from nonartistic visual support. On the other hand, correlational studies show that high aesthetic value of artistic movie sequences perceived by the student is associated significantly with learning enhancement (Bonaiuto et al., 2002). One might wonder how the emotional experience elicited by the appreciation of diverse forms of art enables individuals to feel better and learn quickly and effectively, and whether the boosting effect of art on these different domains forms a basis of a common cognitive or affective mechanism. Here, we suggest that the processing of aesthetic artwork relies on the activity of reward-related brain areas, resulting in positive emotions and pleasure that, modulating affective state, increase the individual predisposition to cognitive activities such as learning. Linking the Brain to Aesthetic Experience The studies reviewed so far demonstrated that the aesthetic value of artwork and their use in educational programs may affect psychological and physiological states, thus promoting well-being and enhancing learning. However, as we stated above, the mechanisms underlying the relationship between art and well-being are still unclear, probably due to the fact that the determinants of the aesthetic experience and its relationship with emotion processing and pleasure are still unresolved. Here, we review some neuroimaging evidence detailing the neural underpinnings of the relationship between aesthetic experience and activation of emotional states in the beholder, to provide a more comprehensive understanding of the aesthetic experience and how it provokes aesthetic emotion and pleasure in the beholder. Moreover, we relate these findings to influential models of aesthetic processing. From a psychological point of view, it has been suggested that the cognitive processing of art produces affective and often positive and pleasing aesthetic experiences. According to the information-processing stage model of aesthetic processing by Leder et al. (2004), the occurrence of aesthetic pleasure depends on a satisfactory cognitive understanding of the artwork. The better the understanding, the more the reduction of ambiguity, and the higher the probability of positive aesthetic emotion. When aesthetic experiences are often positive, it can be expected an increase in positive affect (Leder et al., 2004). Enduring predominance of diffuse positive affective states influences mood (Scherer, 2005), promotes health and learning. Consistently, some neurophysiological studies find that context information facilitates the processing of a work of art and increases positive emotions (Gerger and Leder, 2015;Mastandrea, 2015;Mastandrea and Umiltà, 2016). This is accompanied by greater neural activity in the medial orbitofrontal cortex (OFC) and ventromedial prefrontal cortex, regions strongly associated with the experience of reward and emotion processing (Kawabata and Zeki, 2004;Kirk et al., 2009). On the other hand, various theories of emotion have been influential in describing the paradoxical enjoyment of negative emotions in art (Juslin, 2013;Sachs et al., 2015;Menninghaus et al., 2017). Several authors suggested that the psychological distance of the perceiver from what is depicted in the artwork-which comes from the individual's awareness that the represented object or event is a cultural artifact-reduces the emotional impact of the eliciting object or event and allows the appraisal of the aesthetic qualities of the artwork. This "psychological distance" account underpins the difference between art-specific emotions and utilitarian emotions (Frijda, 1988;Scherer, 2005). Perceiving safety during art reception allows negative content of the artwork to be embraced. In this account, negative emotions such as sadness and sorrow are transformed in source of pleasure and empathetic responses to the emotional content of the artwork are allowed by the meta-emotional reappraisal (Menninghaus et al., 2017). Accordingly, art context influenced aesthetic judgment and emotional responses as measured by facial electromyography (EMG). Specifically, defining visual stimuli as artistic prompted participants to judge artworks depicting negative emotional content more positively, meaning "liked" more. In other words, there might be a general positive bias in the perception of art (Gerger et al., 2014). The pleasurable effect of negative emotions in art reception has been extensively investigated in the field of music (Vuoskoski et al., 2012;Juslin, 2013;Kawakami et al., 2013;Taruffi and Koelsch, 2014;Sachs et al., 2015). According to the BRECVEMA model elaborated by Juslin (2013), enjoying sadness in music derives from the combination of two key mechanisms, i.e., emotional contagion and aesthetic judgment that generate mixed affective responses. While listening to sad music, one may experience the feeling of sadness through the mechanism of emotion contagion and appreciate the beauty of the piece by judging it aesthetically positive (Juslin, 2013). Some authors described the beneficial effects of music listening on the emotional health, reporting that listeners use music to enhance positive emotions and regulate negative emotions, affecting mood (Taruffi and Koelsch, 2014;Sakka and Juslin, 2018). Consistently, an influential model by Sachs et al. (2015) posits that pleasure in response to sad music is functional to restore homeostatic equilibrium that promotes optimal functioning. For instance, a person who is experiencing emotional distress and has an absorptive personality will find pleasure in listening to sad music because, being focused on the aesthetic experience of appreciating the beauty of music will disengage him/her from distress, promoting positive mood. This concept is supported Frontiers in Psychology \| www.frontiersin.org by the fact that listening to sad music engages the same network of structures in the brain (i.e., the OFC, the nucleus accumbens, insula, and cingulate) that are known to be involved in processing other stimuli with homeostatic value, such as those associated with food, sex, and attachment (Berridge and Kringelbach, 2015;Sachs et al., 2015). In line with the conceptual frameworks offered by music research, it may be hypothesized that pleasure in visual art reception relies upon (1) emotional contagion with the valence conveyed by the artwork; (2) appraising a negative emotional stimulus as a fictional rather than realistic; (3) regulating emotion accordingly; (4) enjoying aesthetic experience and performing aesthetic judgment. If aesthetically pleasing, such an experience can be defined rewarding. The dynamic interaction of these and other factors for producing pleasurable aesthetic experience has been broadly described in theories of aesthetic processing (e.g., Sachs et al., 2015;Menninghaus et al., 2017;Pelowski et al., 2017). Providing a comprehensive account of this complex process is out of the scope of this review; however, here we focus on how a part of these mechanisms-i.e., emotion contagion, emotion regulation, pleasure, and rewardfind a common neural substrate in network of emotion processing and how coupling neuroimaging research with measurement of physiological states may be useful for demonstrating a link between aesthetic experience and promotion of well-being. Neuroaesthetics is a relatively recent research field within cognitive neuroscience and refers to the study of neural correlates of aesthetic experience of beauty, particularly in visual art (Chatterjee and Vartanian, 2016). Using multimodal neuroimaging techniques, such as functional magnetic resonance (fMRI), magnetoencephalography (MEG), and electroencephalography (EEG), it has produced heterogeneous results. Most studies, however, converge in the consideration of the orbitofrontal cortex (OFC), and more generally, the core centers of emotional and reward-related responses as the putative correlates of the aesthetic experience of beauty (Kawabata and Zeki, 2004;Di Dio and Gallese, 2009;Ishizu and Zeki, 2013), hence supporting psychological studies that suggest that aesthetic experience is emotionally positive and rewarding (Leder et al., 2004). Using fMRI, it has been shown that rating the beauty of an artwork selectively engaged regions within the OFC irrespective of stimulus type (i.e., visual art, visual texture, music, mathematical formulae, moral judgment etc.) (Blood et al., 1999;Kawabata and Zeki, 2004;Tsukiura and Cabeza, 2011;Jacobs et al., 2012;Zeki et al., 2014). Moreover, metabolic activity in those areas increased linearly as a function of aesthetic, but not perceptual judgment of paintings (Ishizu and Zeki, 2013), indicating that aesthetic preference for paintings is mediated by activity within the reward-related network. Similarly, using MEG to record evoked potentials while viewing images of artworks and photographs, Cela-Conde et al. (2004) found that the left dorsolateral prefrontal cortex (DLPFC) responded more when participants judged the images as beautiful, than when they judged the images as not beautiful (Cela-Conde et al., 2004). Interestingly, Vartanian and Goel (2004) highlighted different neural patterns of activation for pleasant and unpleasant paintings. Specifically, they found that bilateral occipital gyri and left cingulate sulcus activated more in response to preferred stimuli, whereas activation in the right caudate nucleus decreased in response to decreasing preference ratings (Vartanian and Goel, 2004). As activity in the caudate nuclei have been found to decrease following a punishment feedback (Delgado et al., 2000), it may be suggested that deactivation of left caudate reflects a general pattern of reduced activation to less rewarding stimuli (Vartanian and Goel, 2004). In line with these findings, a recent study of Ishizu and Zeki (2017) showed that images rated as beautiful but evoking opposite emotions (i.e., joy vs. sorrow) modulated activity in OFC, but also activated areas that have been found to be involved in positive emotional states (i.e., controlling empathy toward other)-such as the temporoparietal junction (TPJ) and the supramarginal gyrus (SMG)-and negative emotional states (i.e., perception of social pain)-such as the inferior parietal lobule (IPL) and the middle frontal gyrus (MFG) (Ishizu and Zeki, 2017). Consistent with these findings, theories of embodied cognition suggested that emotions may be conveyed by the work of art through embodied simulation (Freedberg and Gallese, 2007;Azevedo and Tsakiris, 2017) or motor contagion (Gerger et al., 2018). In support of this, neuroimaging studies found the aesthetic judgment of human and nature content paintings to be modulated by the activation of a motor component. That is, cortical motor systems were activated including parietal and premotor areas (Di Dio et al., 2015). This suggests that dynamic artworks may engage motor systems via features that represent actions and emotions (Freedberg and Gallese, 2007). Therefore, experiencing art is a self-rewarding activity, irrespective of the emotional content of the artwork. This finding is supported by previous research showing that an art context heightens positive response toward images with negative content (Gerger et al., 2014). Adopting a distanced perspective in art reception may produce positive emotional state and pleasure, irrespective of the emotional content of the artwork (Leder et al., 2004;Menninghaus et al., 2017). Moreover, it appears that art-specific emotions and utilitarian emotions found a common neural substrate in brain network involved in emotion processing and reward. Aesthetic Emotion and Well-Being: Which Relationship? The studies reviewed so far suggest that the aesthetic processing of an artwork can elicit in the beholder affective states congruent with those evoked by the artwork itself (Freedberg and Gallese, 2007;Azevedo and Tsakiris, 2017;Ishizu and Zeki, 2017). Critically, the positive or negative valence of the aesthetic emotion does not appear to be relevant in determining the reward value of the aesthetic experience. A portrait, a sculpture, or a piece of music conveying feelings of sadness may be rated as beautiful and produce a modulation onto OFC regions and the centers of reward-related responses similar to artworks conveying positive feelings, such as joy and pleasure. These results support the claim that adopting a psychological distance Frontiers in Psychology \| www.frontiersin.org in art context allows the perceiver to embrace the negative content of the work of art and, by means of empathetic responses to the content of the artworks, provoking aesthetic pleasure (Menninghaus et al., 2017). According to Marković (2012), the aesthetic experience is an exceptional state of mind, which opposes everyday, pragmatic experience and "protects" the individual from the effects of oppressive reality (Marković, 2012). Given these considerations, it may be thought that the aesthetic emotion is distinctive of aesthetic appreciation, denoting an art-specific emotional response evolved from basic biologic emotions (Leder et al., 2004). As such, this self-rewarding nature of aesthetic experience may account for aesthetic appreciation's promotion of health and well-being. Alternatively, it may be that experiencing positive aesthetic emotions is not only the outcome of a special empathetic state provoked by the artwork but may depend on the level of perceived ambiguity in the artwork itself. In processing fluency theory of beauty, the more fluently the perceiver can process an object, the more positive the aesthetic response (Reber et al., 2004). In other words, features that facilitate processing of a stimulus (e.g., objective stimulus properties and subjective previous experience with the stimulus) result in positive affective responses and more favorable judgments or preferences (Reber et al., 2004). In this view, positive valence of the aesthetic emotion is the product of the processing experience of the perceiver, aesthetic or not. Therefore, aesthetic pleasure can depend, in turn, on satisfactory mastering the stimulus, emotional responses or both (Mastandrea et al., 2009;Chirumbolo et al., 2014). As reviewed above, theoretical frameworks explaining the paradox of enjoying negative emotions in art indicated that different key factors interact to produce a pleasurable response (Juslin, 2013;Menninghaus et al., 2017), as a function of restoring homeostatic balance (Sachs et al., 2015). Intriguingly, the positive affective state derived from the aesthetic emotion, whatsoever origin it may have had, may have a common neural substrate in the reward-related brain circuitry. Nevertheless, these different approaches to aesthetic evaluation may have different implications for a strategic use of art as tool for promoting well-being and health. Consistent with the fluency processing theory of beauty, representational paintings should be more effective than abstract paintings for enhancing learning processes within art-based education programs. Similarly, artwork high in comprehensibility should render healthcare settings or work environments more gratifying than less intelligible artwork. On the other hand, it is possible that experiencing an abstract modern painting in an art museum (i.e., an art context soliciting the adoption of a distanced perspective in the perception of art) can arouse a powerful aesthetic emotion. This could improve perceived well-being (Freedberg and Gallese, 2007;Gerger et al., 2014Gerger et al., , 2018Menninghaus et al., 2017). Unfortunately, as far as we know, there are only a few studies that explore the neural correlates associated with cognitive-or affective-based accounts of the aesthetic experience and their relation to the use of art for promoting individual well-being. Furthermore, most empirical investigations of the relationship between art and well-being do not consider objective measures of stress, such as skin conductance, heart rate variability, or respiration rate. Further, any conclusion about a relationship between art appreciation and well-being is hampered by the use of quite different subjective measures of well-being, such as interviews and questionnaires. Nowadays, we know from the literature that the pleasure associated with aesthetic processing may be modulated by emotional responses of the beholder to the artwork, or may be function of the successful cognitive mastery of the aesthetic stimulus (Leder et al., 2004;Menninghaus et al., 2017;Gerger et al., 2018), or may be a function of a more complex model. Deeper understanding of the dynamic relationship between bottom-up stimulus properties and top-down cognitive appraisal on emotional experience during the aesthetic appreciation of an artwork might be useful to effective use of art-based tools for promoting individual health and well-being. Investigating the interplay between art and well-being must not omit consideration of the analysis of more objective psychophysiological measures of stress, such as autonomic responses. Future research should address the relationship between the emotional responses to aesthetic and non-aesthetic stimuli and measures of well-being, such as combining neural responses with autonomic indices of stress. CONCLUSION Aesthetic experience, in many settings, may promote well-being. Neuroaesthetics research suggests that aesthetic pleasure is derived by the interaction between emotion processing that involves reward-related areas in the brain and top-down processes derived from the relationship of the beholder with the cultural artifact. The self-rewarding nature of aesthetic experience may influence the beholder's affective state, possibly improving wellbeing. However, there still are many questions that future research should address to clarify the determinants of aesthetic pleasure and their relationship to health. First, the impact of aesthetic emotion on measured well-being has been assessed through subjective ratings using interviews or questionnaires, scarcely considering more objective indices recorded through psychophysiological measures. Moreover, it remains unclear whether proper use of art to improve well-being should emphasize the empathetic responses to the artwork or the possibility for the beholder to master the meaning of the artwork itself. Future research should consider these issues in developing art-based programs in healthcare and education. AUTHOR CONTRIBUTIONS SM conceived the idea, reviewed the literature, and wrote the draft of the manuscript. SF reviewed the literature and wrote the draft of the manuscript. VB collaborated to the idea with SM, reviewed the literature on educational applications and supervised the manuscript writing.	2019-04-04T13:07:13.995Z	2019-04-04T00:00:00.000	{ "year": 2019, "sha1": "383a9db75684801c8a9f94b08eee275b76a9e561", "oa_license": "CCBY", "oa_url": "https://www.frontiersin.org/articles/10.3389/fpsyg.2019.00739/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "383a9db75684801c8a9f94b08eee275b76a9e561", "s2fieldsofstudy": [ "Art", "Psychology" ], "extfieldsofstudy": [ "Medicine", "Psychology" ] }
150353889	pes2o/s2orc	v3-fos-license	The Patient Teacher in General Practice Training: Perspectives of Residents Background: Patient teachers were involved in training general practice residents (GPRs) to strengthen the patient-centered approach. They teach a course on health democracy by themselves and teach in tandem with a physician teacher during reflective practice-based classes (named GEPRIs). We present the GPRs’ representations of patient teacher characteristics and capacities and their perception of how useful patient teachers are to their professional development. Methods: We administered a questionnaire based on a preliminary qualitative study to 124 GPRs. It explored (a) changes in the GPRs’ representations about patient teacher characteristics and capacities with regard to teaching over the first year of the experiment; (b) GPRs’ perception of patient teacher utility to their training and their contribution to developing patient perspective–related competencies. Results: The response rate was 89.5% (111/124). The majority of GPRs agreed with 17 (before) and 21 (after) of the 23 patient teacher characteristics and with 17 (before) and 19 (after) of the 20 capacities. The agreement rate increased, overall, after patient teacher participation. The GPRs found patient teacher useful to their training in 9 of 11 topics (agreement rate 65%-92%). They felt they had developed the 14 patient knowledge–related competencies (agreement rate 62%-93%), and 52% to 75% of the GPRs rated the patient teachers’ contribution to those competencies “high or very high,” depending on the competency. Conclusion: This study demonstrates the specific contribution of patient teachers to university-level medical training in France. The GPRs recognized that patient teachers helped them develop competencies by providing patient-specific content. Introduction Background Patient involvement in medical training has been on the rise since the 1970s (1), driven by the push for greater social responsibility on the part of medical schools, as promoted by the World Health Organization (2). More recently, a number of reports have stressed the importance of patients' participation as members of the teaching team for education, evaluation, and research on this educational innovation (3,4). Several reviews and syntheses of the literature (1,(5)(6)(7)(8)(9) have highlighted the different forms and levels of patient involvement in the health system. In the context of teaching, they also studied the benefits perceived by students (improvement in technical skill, communication, collaborative skills, etc), by teachers and professionals (recognition of the patient perspective), and by patients (raised self-esteem and empowerment) (1,8). In particular, it was noted that patients could be considered true colleagues in medical education as long as they are supported, trained, and paid. Towle et al stressed in 2010 (1) the importance of supporting coordinated programs involving patients as authentic institutional partners for both teaching and curriculum assessment and development. Although experiences have been implemented and evaluated in various countries (mainly in the United States and United Kingdom) (8), the prolonged involvement of patients at an institutional level in the training of French general practitioners is very recent and its acceptability by the residents is questioned. In France, health democracy-introduced by Law No. 2002-303, adopted March 4, 2002, regarding the rights of sick persons and the quality of the health-care systemfacilitated and encouraged more meaningful patient involvement in the health-care system. As a result, users and patient organizations were gradually incorporated at all levels of the health-care system, based on their experience (10). This was seen as a driver of health-care system quality (11). It was in this context that a General Practice Department at a medical faculty of a French university launched a training program in 2016 with patients as teachers (12). Presentation of the French General Practice Resident Training Program After their first 6 years of study, medical students pursue a specialist degree, called a diploˆme d'e´tudes spe´cialise´es. Teaching takes the form of classroom instruction and reflective practice groups (named GEPRIs) aimed at developing critical thinking skills, in accordance with cognitivist and constructivist learning theories (13). During a GEPRI, 3 general practice residents (GPRs) present and analyze clinical situations they have encountered during their residency in a small group setting. With the aid of the instructors, the group analyzes the clinical, psychological, and social aspects of those situations. After the GEPRI, the 3 residents write what is called a Re´cit de Situation Complexe Authentique using the group analysis and add a literature review. This includes a description and analysis of the situation, an analysis of the decisions made, a definition of the problems, the responses to those problems, and then a recontextualization via a critique of the decisions made and the construction of an optimal strategy. At the end of the GEPRI, the group chooses two of the problems that emerged in the course of the discussion, and 2 residents give an oral presentation on those problems at the next session. By the end of the 3year program, each GPR will have participated in 21 GEPRIs. Patient teachers were introduced as part of the general practice training program. That innovative program had its foundations in the drive for social responsibility in primary care (14) and in the reform of GPR's training, which made the "patient-centered approach" central to the competencies needed by future general practitioners (15). Programs based on that view have been shown to improve students' ability to apply the patient-centered approach (16). To foster acceptance and implementation of its program, the General Practice Department relied on existing recommendations (1,3). Patient coordinators are members of the teaching committee. They are charged with recruiting patients as teachers, using a grid based on criteria set out in the literature (7). The patients recruited are volunteers, belong to a patient organization, and have good health knowledge and experiential knowledge (17); they have a clear idea about what they want to convey to the GPRs; and have good will toward GPRs (18). That choice of patient profile was based on the assumption that such patients would have a quality-of-care-oriented perspective and the same critical reading of the situations being analyzed. The 20 patient teachers recruited have the status of casual teachers. They are paid. They help design the lessons, teach the first class on health democracy on their own, and teach in tandem with a physician teacher in 80% of the GPRs' classes. They bring a stronger patient perspective to residents' analyses and practices by responding to what the residents say they know about patient expectations and by offering health resources from organizational settings with which the residents are not familiar (18). Lastly, patient teachers participate in the educational assessment of the GPRs. Program development is supported by a steering committee made up of patient teachers, physician teachers, and researchers, in accordance with the design-based research approach (19). Two studies accompanied the program implementation. They first analyzed the content of patient teaching-identified based on observation-as it related to the competencies expected of general practitioners (18). As both the patients and physician teachers accepted the ethical and political aspects of patient involvement in medical teaching (20), the second study examined the residents' opinions about patient teacher participation. This article presented the result of this second study. Methods Sample One hundred twenty-four GPRs from 2 (second-and thirdyear) cohorts at University Paris 13 were asked their opinion about patient teacher involvement in their training. Materials As the literature offered no questionnaires that could be used to document residents' representations regarding patient teacher involvement, the first phase consisted of a qualitative study among the stakeholders prior to participation in the program (21). There were semistructured interviews with 7 patient teachers and 8 physician teachers and focus groups with 9 GPRs at the beginning of the program. This provided data on the different stakeholders' representations regarding the characteristics and capacities they felt that patient teachers should possess, as well as on general practitioner's competencies to which patient teachers might usefully contribute. Those results were used to construct a questionnaire. Because-for organizational reasons-patient teacher involvement began early, the questionnaire was not administered until the end of the first year of the experiment. Despite the well-known limitations of this type of survey, we wanted to document how the GPRs' representations regarding required patient teacher characteristics and capacities changed from the beginning to the end of the year-long experiment on a single questionnaire. The questionnaire had 3 main sections, plus questions to identify the population: 1. The first part asked for the GPRs' representations regarding 23 characteristics (Table 1) and 20 capacities (Table 2), before, and then after, being taught by a patient/physician pair. 2. The second part asked their perception of patient teacher utility with regard to 11 topics Table 3). 3. The third part evaluated their sense of competence concerning 14 patient perspective-related competencies and the degree to which patient teachers helped them develop those competencies (Table 4). 4. Identification questions: year of training and the number of reflective practice groups (GEPRI) participated in with a patient teacher. All of the assertions were rated using a 4-point Likerttype scale: strongly disagree, disagree, agree, and strongly agree. The degree of patient teacher contribution to the acquisition of competencies was evaluated using another 4-point scale: none, low, high, and very high. At the end of each section, an open question allowed respondents to add comments. Before being administered, the questionnaire was tested on 5 health-care professional trainers. There were changes in both the presentation of the questionnaire and the formulation of certain assertions. The finalized questionnaire was then validated by the steering committee. The questionnaire was administered electronically (Lime-Survey) to all of the residents from January 15, 2017 to February 12, 2017. Three reminders were sent. Participation was anonymous and voluntary. Analysis We used descriptive statistics to analyze changes in representations regarding patient teacher characteristics and capacities « before and after patient teacher involvement ». The agreement and disagreement rates were calculated considering the percentage of respondents for each assertion. At the implementation stage of our experience, it seemed important to us to better discriminate opinions in agreement or disagreement with the proposals made. In that way, the 4 levels of agreement or disagreement on the Likert scales were reduced to 2: agree ("agree" and "strongly agree") and disagree ("disagree" and "strongly disagree"). The data were compared as a function of the residents' year of study and the number of reflective practice learning groups they had participated in, to verify The open comments underwent an inductive thematic analysis in relation to the study objectives. A researcher codified each comment. Then the comments were classified into 3 main themes, including subthemes: (1) Positive opinions: on the capacities of patient teachers, their specific contributions, their usefulness in the development of certain skills; (2) Negative opinions: the lack of relevance of the patient teacher's intervention during GEPRI, the fear of being judged by patient teachers, the lack of representativeness of patients, differences in attitude according to patient teachers; (3) Suggestions for training. Change in the Residents' Representations on Patient Teacher Characteristics The majority of the GPRs agreed with 17 of the 23 proposed characteristics before and after patient teacher involvement; the rate of agreement ranged from 59% to 98% (Table 1). The percentage agreement increased (by 1%-33%) after patient teacher involvement for 22 of the characteristics. The residents' opinion was unchanged for only 1 characteristic, at 75% agreement. For 6 characteristics, the percentage agreement before patient teacher involvement was less than 50%. For 4 of these, opinions changed after patient teacher involvement to at least 50% agreement, increasing by 5% to 33% agreement, depending on the characteristic. For the other 2 characteristics, the agreement rate remained low, at less than 50%. These were "The patient teacher has training in teaching" and "The patient teacher participates in accordance with the same educational values as the physician educators." Change in the Residents' Representations on Patient Teacher Capacities The majority of the residents agreed with 17 of the 20 proposed capacities before and after patient teacher involvement; the rate of agreement ranged from 54% to 93% (Table 2). Before patient teacher involvement, 3 patient teacher capacities received less than 50% agreement: the ability to convey scientific knowledge about his illness (43%), to participate in assessing residents (39%), and to teach on their own (33%). Agreement on 2 of these capacities increased to over 50% after patient teacher involvement (by 13% and 14%). The majority of GPRs continued to disagree that the patient teacher was capable of teaching on his own in the general practice training program (33% agreement before and 39% after). Agreement with 16 of 20 capacities increased after patient teacher participation (by 2%-18%). The GPRs' opinion was unchanged after patient teacher participation for 3 capacities. One capacity received slightly less agreement "after" (74% before vs 71% after): the patient teacher's ability to adapt his participation to the residents' behavior and attitudes. Perception of Patient Teacher Utility in the Residents' Training Program The GPRs found the patient teachers useful in the training program for 9 of the 11 proposed topics (65%-92% agreement). They were seen as less useful for 2 topics: the patient teacher's usefulness for better understanding the clinical problems presented in the GEPRIs (37% agreement) and for acquiring new medical skills (41% agreement; Table 3). Competency Acquisition and Patient Teacher Contribution to Competency Development The majority of GPRs felt they had developed the 14 patient perspective-related competencies (62%-93%). For 13 of the 14 competencies, they felt that the patient teacher's contribution to developing those competencies was either high or very high (52%-75%; Table 4). For the "taking the entourage (peer helpers) into account when making medical decisions" competency, a slight majority (53%) felt that patient teacher contribution was low. Comparison of Responses by Respondent Profile For some patient teacher characteristics and capacities, the agreement rate before patient teacher involvement was higher among second-year residents than among third-year residents. Second-year residents were more likely to consider patient teachers experts in their illness (P ¼ .04), capable of staying in their role as patients when conveying their personal point of view (P ¼ .01), and capable of adapting their contribution to the residents' behavior and attitudes (P ¼ .05). After patient teacher involvement, the secondyear residents were more likely to find them useful for better understanding the reasons for treatment noncompliance (P ¼ .02) and more likely to feel that they helped them develop the "conducting a shared decision-making process with the patient" competency (P ¼ .01). The third-year residents, on the other hand, were more likely to feel capable of helping patients develop the competencies they need to manage their illness (P ¼ .02). Open Comments Fifty (46.8%) of 111 GPRs offered 115 comments (between 1 and 3 comments per resident). A total of 101 usable comments were analyzed. Positive Opinions Fifty-four percent of the comments (offered by 24/50 GPRs) reflected a positive view of patient teachers by GPRs and the usefulness of incorporating the "patient perspective" into general practice training. They stressed the qualities of the patient teachers ("committed," "involved," "qualified," "very receptive", etc). They welcomed the patients' perspective because it helped give a broader, more complete view of the experience of the illness and what patients feel ("not just the emotional experience, but also the material, financial, and professional circumstances"). The residents felt that patient teacher involvement would improve the care relationship, thanks in part to a better understanding of the patient ("very useful for understanding our patients with chronic illnesses"), not just by understanding what they feel but also by becoming more conscious of their own attitudes ("better grasp the attitudes that we can adopt," "helps us know the impact of what is said and done by doctors," and "makes us more humble in how we treat patients"). Some comments mentioned specific supplementary patient teacher contributions ("nonscientific information . . . the entire social, organizational, health economics side," "information on the health-care system, the patient care pathway, their rights, legislation," and "resources that aren't taught in our general practice curriculum"). Negative Opinions Fourteen residents expressed varying degrees of reticence about patient teacher involvement in training GPRs (32% of the comments), ranging from complete rejection to the expression of a variety of limitations. The latter had to do with the status of patient, which seemed to them incompatible with teaching medicine ("because I don't think people who don't practice medicine can understand the problems we encounter"). Being an expert in only one illness was seen as a limitation ("it's hard for them to have knowledge about other illnesses"). The presence of patient teachers in reflective practice groups made some residents feel judged, attacked, or less secure ("not a place where we have to explain ourselves or defend ourselves over what went wrong with patients, but rather to find ways to improve our doctorpatient relationships without feeling guilty"). Lastly, some complained about certain patient teacher attitudes ("too activist, demanding, and aggressive in their comments"). Suggestions for Training Twelve residents offered suggestions. Some felt that it was not always useful to have a patient teacher at all of the GEPRIs. They suggested practical workshops with patient teachers for working on communication skills (eg, "their presence should probably be used more for trying out our communication skills and our ability to create a therapeutic relationship"). Three residents mentioned the possibility of having other health-care professionals participate as well. Discussion Even assuming that the residents were able to remember their initial representations regarding patient teacher characteristics and capacities, this discussion places the primary emphasis on the results obtained at the end of the yearlong experiment. This study shows that GPRs have a positive view of patient teacher involvement in their general practice training. They identify the patient teachers' characteristics and capacities and the latter's usefulness in developing certain general practice competencies, helping the GPRs incorporate the patient perspective into their practice. These results confirm and add to those in the literature on patient participation in medical training (1,7). This study also shows that beyond patient teachers' personal narratives (22) and their role in teaching clinical and communications skills (1), patient teachers can be involved in learning processes such as those employed in reflective practice groups. An increasing number of health-care training curricula include reflective practice training, which significantly enhances students' learning and helps them better incorporate theory into their practice (23). The 2009 literature review reported a single qualitative study in which mental health patients participated in nursing education based on reflective practice classroom (8). In our study, patient teachers make a real contribution to residents' reflective analysis, especially since such learning is seen as a positive experience and occurs in a supportive learning environment (24). The residents recognize the special status of patient teachers (someone with a chronic condition, belonging to an organizational network, activist, user representative, etc), their special knowledge (of the laws and the health-care system, scientific and experiential knowledge of the disease, etc), and their motivation for teaching and improving both preventive and curative care. These characteristics are consistent with those used to recruit patient teachers for this project, confirming their validity inasmuch as they enable patient teachers to create cognitive conflict, which facilitates incorporation of the patient perspective by the GPRs (13). Hence, this study underlines the specific contributions of patient teachers recruited using clearly defined criteria in the training of GPRs-contributions that physician teachers would be unlikely to provide on their own or would approach differently (25). A comparison between the second-and third-year residents' responses indicates better acceptance of patient teachers by the second-year residents on some items. One possible hypothesis is that second-year residents still consider themselves students, susceptible to being thrown off balance, unlike third-year residents who-soon to be in practicewant to feel sure about what should be done. Note, however, that unlike the third-year residents, the second-year residents experienced patient teacher involvement right from the start of their program. This might indicate better acceptance when patient teachers are introduced early in the residents' training-something worth verifying in a study. There was some reticence about patient involvement in reflective practice groups. In contrast to other studies in which medical students felt less intimidated by the patients than by a physician instructor (26,27), a few residents felt that patient participation in the reflective practice groups was threatening to their learning. This is explained, in part, by the type of involvement that practice analysis demands and shows that the framework and rules for participation must be clearly defined for everyone at the start of each session. Some authors have mentioned things likely to have a negative impact on reflective practice, such as an intellectually or emotionally nonconducive environment or one in which opinions cannot be expressed freely (23). The GPRs questioned the idea of patient teacher involvement in activities other than cofacilitating reflective practice groups; the majority (61%) did not consider it desirable to have patient teachers teaching on their own in the general practice curriculum, and nearly half (49%) felt that patient teachers should not participate in assessing GPRs. Nevertheless, the program does intend to have patient teachers taking part in the educational assessment of residents. It will therefore be important, in future studies, to look at the specific patient teacher contributions to such assessments and to evaluate their perceived utility among the GPRs (28,29). Unlike in other countries, patient involvement in medical training is relatively new in France, which might explain the reticence observed. In addition, some of the open comments suggested that patient teachers be involved in other forms of practical teaching-for example, in helping them improve their communication or clinical skills-or even with other health professionals, as described in the literature (1,30,31). Although a group created specifically for the new program coached the patient teachers throughout their involvement, this initial experiment highlights the importance of preparing patient teachers-as well as physician teachers-for teaching. Indeed, the 2-person approach to teaching and facilitating practice analysis groups requires specific skills that can be acquired by either formal or experiential training (32). For this project, the patient teachers were trained via patient meetings (roughly every 2 months), under the guidance of patient researchers, as recommended in the literature (25). The content was harmonized at those meetings by pooling the patient teachers' experiences and approaches. The meetings gave the patients an opportunity to reflect on their involvement and construct broader educational messages (not limited to their own pathology). Study Limitations This study had some limitations. The opinions of the GPRs on patient teacher's characteristics and capacities both "before and after patient teacher's involvement" were collected at the same time, after patient teacher's involvement. This constitutes bias. General practice residents' opinion "before patient teacher involvement" may have been colored by their exposure to the patient teacher(s) during their training (reflective practice groups and lessons on health democracy). Another limitation is related to the use of an invalidated questionnaire in the absence of a questionnaire identified in the scientific literature. Some of the questions proved to be poorly formulated or imprecise, making interpretation difficult. In the future, the improved questionnaire will be administered at the beginning and end of each year of training in order to confirm these results. The qualitative analysis based on GPR's comments may also constitute a bias by the fact that only one researcher conducted this analysis, diminishing the reliability criterion specific to qualitative research. Lastly, the results obtained cannot be generalized directly to all university general practitioner medical courses. Indeed, each training course differs from the others by its pedagogical approach. Our study looked at students' perceptions on the place and contribution of patient teachers during the reflective practice groups, an educational method not found in all training courses. In fact, the results can only be generalized if they are discussed according to the pedagogical formats of the training and the determined role of the patient teachers. Conclusion This study demonstrated the utility and-despite some dissenting voices among the GPRs-recognition of patient teachers' specific contributions to general practice training in France. Patient teacher participation-along with physician teachers-in reflective practice groups is new, and the residents found the experience to be positive and useful to their learning process. The GPRs felt that the patient teachers helped develop their competencies by providing patient-specific content. This innovative program is continuing with new student cohorts. To complement these results, it would be interesting to assess the longer term impact of patient teacher involvement on general practitioners. Consistent with patient engagement in health professional's training, the perspective of health professionals is essential to improve the training program. Similarly, the participation of patients in the organization of the curriculum is important as soon as their status is recognized and valued financially. Authors' Note M.J.A. conducted the study, including designing the questionnaire, collecting and analysing the data, and drafting the manuscript. R.G. and O.G. helped design the study (questionnaire development), interpret the results, and improve the manuscript. C-.A.K. and S.H. helped design and administer the questionnaire, collect the data, and improve the manuscript. A.M. and Y.R. helped design the study, interpret the results, and improve the manuscript. C.M. supervised all phases of the study, including questionnaire design, data analysis, and manuscript development. All of the authors participated in manuscript preparation and have approved the version submitted for publication.	2019-05-12T14:23:50.562Z	2018-10-02T00:00:00.000	{ "year": 2018, "sha1": "c90daaebafb47f58fe4cd59001d221de1de65407", "oa_license": "CCBYNC", "oa_url": "https://journals.sagepub.com/doi/pdf/10.1177/2374373518803630", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "7583325d2bc46cad491a8ce7cd2a5fa90a2ae850", "s2fieldsofstudy": [ "Medicine", "Education" ], "extfieldsofstudy": [ "Psychology", "Medicine" ] }
59618613	pes2o/s2orc	v3-fos-license	Learning from the transfer of a fellowship programme to support primary care workforce needs in the UK: a qualitative study Objectives Service redesign, including workforce development, is being championed by UK health service policy. It is allowing new opportunities to enhance the roles of staff and encourage multiprofessional portfolio working. New models of working are emerging, but there has been little research into how innovative programmes are transferred to and taken up by different areas. This study investigates the transferability of a 1-year post-Certification of Completion of Training fellowship in urgent and acute care from a pilot in the West Midlands region of England to London and the South East. Design A qualitative study using semistructured interviews supplemented by observational data of fellows’ clinical and academic activities. Data were analysed using a thematic framework approach. Setting and participants Two cohorts of fellows (15 in total) along with key stakeholders, mentors, tutors and host organisations in London and the South East (LaSE). Fellows had placements in primary and secondary care settings (general practice, emergency department, ambulatory care, urgent care and rapid response teams), together with academic training. Results Seventy-six interviews were completed with 50 participants, with observations in eight clinical placements and two academic sessions. The overall structure of the West Midlands programme was retained and the core learning outcomes adopted in LaSE. Three fundamental adaptations were evident: broadening the programme to include multiprofessional fellows, changes to the funding model and the impact that had on clinical placements. These were felt to be key to its adoption and longer-term sustainability. Conclusion The evaluation demonstrates a model of training that is adaptable and transferable between National Health Service regions, taking account of changing national and regional circumstances, and has the potential to be rolled out widely. Objectives Service redesign, including workforce development, is being championed by UK health service policy. It is allowing new opportunities to enhance the roles of staff and encourage multiprofessional portfolio working. New models of working are emerging, but there has been little research into how innovative programmes are transferred to and taken up by different areas. This study investigates the transferability of a 1-year post-Certification of Completion of Training fellowship in urgent and acute care from a pilot in the West Midlands region of England to London and the South East. Design A qualitative study using semistructured interviews supplemented by observational data of fellows' clinical and academic activities. Data were analysed using a thematic framework approach. setting and participants Two cohorts of fellows (15 in total) along with key stakeholders, mentors, tutors and host organisations in London and the South East (LaSE). Fellows had placements in primary and secondary care settings (general practice, emergency department, ambulatory care, urgent care and rapid response teams), together with academic training. results Seventy-six interviews were completed with 50 participants, with observations in eight clinical placements and two academic sessions. The overall structure of the West Midlands programme was retained and the core learning outcomes adopted in LaSE. Three fundamental adaptations were evident: broadening the programme to include multiprofessional fellows, changes to the funding model and the impact that had on clinical placements. These were felt to be key to its adoption and longer-term sustainability. Conclusion The evaluation demonstrates a model of training that is adaptable and transferable between National Health Service regions, taking account of changing national and regional circumstances, and has the potential to be rolled out widely. IntrODuCtIOn UK health service policy is looking to service redesign as a way of addressing the challenges facing the National Health Service (NHS). [1][2][3] Within primary care, training initiatives (including additional training in hard to recruit posts, the development of portfolio roles for both newly qualified staff and those reaching the end of their careers and workforce development in teams wider than general practitioners (GPs)) are suggested as ways of enhancing the roles of staff, including nurses. 3 4 This has included funding for 250 post-Certification of Completion of Training (CCT) training posts in England, targeted at areas with the poorest GP recruitment, to enable GPs to access additional training in a specialism of interest while addressing local need. 4 Such initiatives are important at a time when the number of GPs intending to reduce their hours or leave general practice is rising in the face of increasing workload. 5 6 They offer experience (cross-sector working, skills enhancement including leadership and management training, and clinical skills training) that goes further than that included in the current 3-year vocational training schemes. 7 8 This mirrors the expanding remit of general practice, with recognition that strengths and limitations of this study ► Few studies have evaluated the delivery of new training programmes for general practitioners and primary care professionals in terms of their transferability from one area to another. ► This study evaluated an innovative additional year of training and had a high level of participation from the cohort eligible for inclusion, with their perspectives gathered at a number of stages of the programme. ► By including a wide range of individuals who worked with the fellows including stakeholders, host organisation leads and colleagues, the study gained a broad perspective of the adoption of the fellowship programme and factors that influenced its transferability. ► Although limited to two regions, together these cover 31.8% of the population of the country and two of the four Local Education Training Boards in England, so strengthening the generalisability of the findings. Open access traditional models of training and continuing professional development in general practice are no longer sufficient to prepare individuals for roles that cross boundaries of care. 9 10 Uptake of service innovation within the NHS is known to be slow with few formal mechanisms existing for spreading learning across services or different geographical areas. 11 Within primary care, evidence suggests the fit between the innovation and the local context is crucial if implementation is to be successful. 12 Where innovation has been shown to be successful, there has consistently been strong leadership or champion buy in and appropriate funding alongside perceived external and internal need. [12][13][14][15][16] Much of the evidence that does exist focuses on facilitators and barriers to innovation with less evidence of how and why some are successful. 12 We recently reported an evaluation of a 1-year post-CCT fellowship programme, developed and piloted by Health Education England (HEE), West Midlands, that provided recently trained GPs with advanced skills training in urgent and acute care, leadership and academic practice. 17 Details of the fellowship programme are shown in box 1. Although positively evaluated, questions remained over scalability and transferability to more complex health service settings. 17 In 2016, HEE, London and the South East (LaSE) adopted the West Midlands fellowship programme throughout the region, so creating an opportunity to study its transferability to multiple contrasting areas. Whereas the secondary care-based elements of the West Midlands' pilot were located in relatively small county hospitals, the LaSE scheme included large inner city hospitals in socially diverse settings. Hence, the aims of this evaluation were to consider the transferability and implementation of the fellowship scheme, in particular looking at how and why it evolved, in order to draw out implications for the further roll-out of such workforce initiatives. MethODs This qualitative study comprised interviews with key individuals, along with observations of fellows in a cross section of workplace settings, to gain in-depth understanding of views and experiences relating to the transfer of a workforce programme from one setting to another. recruitment and data collection All fellows in each of two cohorts of the 1-year urgent/ emergency care fellowship programme implemented in LaSE in 2016 were invited to take part in the study, along with their mentors and key individuals they identified in each of their clinical placements. In addition, we invited key stakeholders responsible for the implementation of the programme, including HEE primary care leads, quality and performance managers and academic leads. All eligible individuals received written study information and were verbally consented. They were also informed what the data would be used for and that confidentiality would be assured. All data were anonymised with unique identifiers assigned to each participant according to the group to which they belonged (HEE=stakeholders and Health Education England staff members, M=fellow's mentor, F=fellow and H=key individual in the healthcare provider organisation). Semistructured interviews were conducted face-to-face or over the telephone and lasted between 20 and 45 min. Initial interviews, conducted around 6 months into the fellowship, explored interviewee's aims, expectations and experiences of the fellowship programme. Second interviews were conducted on or after completion of the programme and focused on the overall experience of the fellowship and its impact on career plans (fellows) and organisational impacts, including capacity building (stakeholders and hosts). Observations of fellows (10 in total) were pragmatically chosen to cover all primary and secondary care settings in which the fellows were hosted, as well as academic days, and to minimise disruption to clinical teams. An observation checklist was used to record evidence of teamwork, box 1 Aims and structure of fellowship programme in West Midlands Seven general practitioners (GPs) within 3 years of post-Certification of Completion of Training participated in the programme in the West Midlands Aims ► To enhance the skills and experience of GPs in urgent/emergency care teams ► To enable GPs to apply enhanced urgent and acute skills to support the development of alternative community-based care pathways ► To raise GP interest in hybrid emergency/urgent and primary care roles ► To support the national policy drive for integration of primary, secondary and social care Programme structure ► 40% time in primary care: GP training practice ► 40% time in clinical attachments: three attachments each of 4 months' duration comprising emergency department, a medical admissions unit and an ambulance service ► 20% academic study: undertaking a bespoke postgraduate certificate in urgent and acute care and participation in an action learning set Core learning outcomes ► Demonstrate the ability to diagnose and assess urgent presentations in long-term illnesses ► Formulate, implement and evaluate current pathways of care according to best evidence ► Show understanding of frailty and complex co-morbidities, particularly in the elderly and how such patients are appropriately managed ► Demonstrate competence in the interpretation and evaluation of evidence and the application of appropriate treatment and assessment ► Apply knowledge and skills to the management of urgent care ► Critically interpret and evaluate the current evidence behind urgent care Open access integrated care working, communication across settings, teaching and academic activity. Observations lasted between 4 and 7 hours during which time members of the clinical team with whom they were located were opportunistically asked to participate in short interviews. Data analysis All interviews were recorded, transcribed verbatim, anonymised and checked for accuracy by CB and RR. Analysis was aided by the use of Nvivo V.11 software package. Using a thematic framework approach to interrogate the data and identify key themes, 18 initial codes were deductively drawn from the research questions. Through further reading of the transcripts, we inductively coded for any elements not previously captured. A thematic framework was devised using an iterative process until all the codes had been identified. 18 CB and JD met regularly to discuss the analysis and identification of emergent themes. Illustrative quotes were identified to elucidate each theme. Patient and public involvement Patients and public were not directly involved in this study. results Participants and settings Of 17 eligible fellows, 15 agreed to participate in the evaluation; one had personal circumstances that prevented them from doing so. In addition, 35 stakeholders, provider organisation clinical leads, GP tutors and mentors participated in planned interviews. Twenty participants were involved in a second interview and six were interviewed a total of three times, as shown in table 1, giving a total of 76 interviews. The timing of when data collection occurred, in relation to the employment period of each of the fellows, determined the extent to which they could each be followed up. An additional 27 interviews (lasting between 5 and 15 min) were completed opportunistically during observation sessions. These included members of GP, emergency department, ambulatory care, urgent care and rapid response teams. Table 2 shows the mix of clinical placements that were experienced by the 15 participating fellows. While most had 2 days/week in general practice, the secondary care placements were highly variable and for one fellow included no direct patient contact. Comparison with and learning from the West Midlands pilot Interviewees described a high level of commitment between HEE partners in West Midlands and LaSE to share learning relevant to the transfer of the fellowship programme, particularly during the year prior to the LaSE fellowship launch. Respondents also highlighted the key role that programme champions in LaSE (from regional level to local clinical educators) played in its successful implementation. The overall aims and structure of the West Midlands programme were retained by LaSE (see Box 1). LaSE adopted the same core learning outcomes, adding a further two covering understanding of ambulatory care and working towards admission avoidance strategies. While the West Midlands fellowship programme was administered across one HEE local area, in LaSE it was across four reflecting a more complex and varied administrative landscape. There was evident commitment between HEE partners in West Midlands and LaSE to share learning relevant to the transfer of the fellowship programme. HEE leads had met and discussed how the pilot programme was set up in the West Midlands, and this fed directly into the development of the LaSE programme. We identified three clear areas of adaptation which will now be explored in more detail. Acceptability and experience of the scheme The stakeholders, mentors and hosts in LaSE viewed the programme favourably, stating that they would be willing to host a fellow in the future. I can't praise him [academic mentor] highly enough actually, I think his style as a programme lead has been brilliant. So in terms of the academic days they're very good. (F10) The programme was also felt by most participants to be fulfilling expectations that it was preparing fellows for portfolio careers, including leadership and academic roles It [fellowship] helps in a number of ways. You can apply it to the academic side, you've got the post-graduate certificate. You can apply it to the fact that you've got a range. (F07) the development of a multiprofessional fellowship model While the West Midlands pilot programme only included GPs, at LaSE it was broadened to include advanced nurse practitioners (ANPs) and physician associates (PAs): two ANPs were included in cohort 1, and two ANPs and one PA in cohort 2. Commissioners and the programme team drove this change as they considered multiprofessional working a progressive development: Open access … the model for urgent and emergency care is predicated in the future on a mixed economy of health professionals. (H04) Nursing fellows welcomed the broadening of the scheme as they described a lack of professional development or upskilling opportunities. The teaching element of the programme was seen to be enhanced by the multiprofessional mix: So one advantage of our programme is that we take all comers, not just GPs, and that's been incredibly useful. Certainly I've noticed when teaching the group … a very heterogeneous group is always better to be teaching and working with. (HEE02) Although the multiprofessional mix was generally well-received, there were some concerns raised about the suitability of non-GPs and the available clinical placements in acute settings. Some of the ANP and PA fellows had difficulty in accessing suitable placements, and some of the placement mentors were unsure of how to best use the fellow: This highlights the need for all participants in the scheme to have clear information on the role of the fellows and the programme purpose. While the fellowship programme was not designed to be competency-based, concerns were raised about the experience and qualification levels of the nursing fellows compared with GP fellows. They're so variable, because you just don't know what background they're coming with. So you know, with the GPs traditional training, they've had two years in hospital medicine and a year in general practice. With an ANP, it depends on what the training's been previously. (Mentor 22) Placement difficulties also arose over uncertainty regarding ANPs' indemnity in some settings: … it wasn't even the funding, I think it was the cover, insurance or litigation. I wasn't able to work there. (Fellow 01) Despite these difficulties, including ANPs and PAs in the fellowship programme was generally viewed positively as a means of providing upskilling opportunities, encouraging individuals to pursue more challenging roles and to increase capacity. I think if we can get them to autonomous practising at urgent emergency care level then they are a very, very employable asset. (Mentor 23) Changes to the funding model While the initial pilot of the fellowship programme had been fully funded by HEE West Midlands, in LaSE the funding climate did not allow this and alternative funding mechanisms were needed: In the West Midlands they were paying 100% of the salary of the individuals involved in the fellowship, and we felt that actually that wasn't a model that would be sustainable as we moved forwards. So we devised a different funding model which was a bursary based model which then left the service element to be funded through service providers and clinical commissioners. (HEE05) In LaSE, the academic element of the fellowship continued to be funded through HEE, with the remaining costs of the scheme being funded by the primary and secondary care organisations providing clinical placements. While this enabled the inclusion of a larger number of fellows, it also led to increasing variation in employers' expectations of the fellows. In addition, the complex employment arrangements were time consuming to set up and manage: I've tried to be quite proactive and I've engaged the employers for several months beforehand and tried to make really sure they know what they're offering and whose responsibility is whose. (HEE04) The LaSE programme required a Clinical Commissioning Group (CCG) or a GP federation/partnership to host the fellow and act as their main employer, taking on responsibility to ensure that the fellowship was financially viable, and cross-charging for the time the fellows spent in other clinical settings: If you take on somebody full time in a Fellowship position the salary cost is £100,000 and the Fellowship grant is £30,000, so you have to balance the £70,000 … … So we have to find them projects to do with organisations that are happy for us to cross charge them for their clinical time. (Host 02) This funding model allowed for flexibility, enabling most fellows to build placements around their interests; however, a few fellows cited the necessity of their host to recoup costs as the main reason they lacked the breadth of experience they had envisioned: I feel completely cheated. I feel like I've been used as a commodity … for my year my key aim was to have the clinical side of it, and that hasn't happened and isn't going to. (Fellow 10) Organisational stakeholders considered that host organisations' investment in the programme was central to its relevance and sustainability: Because service is not getting a freebee or a total freebee they are actually committed to ensuring and investing in it to get the right thing for them as well as the programme itself. So it is buy in … it is a model that can then be replicated across the system as it demonstrates that providers recognise that this kind of approach is really important both for developing future leadership service but also demonstrating an integrated approach to service delivery. (HEE05) Clinical placement experience Although the programme in LaSE retained the same 40:40:20 proportions as in the West Midlands scheme (see Box 1), the organisation of clinical placements differed. In the West Midlands' pilot fellows worked in one GP practice and rotated through three service placements, each lasting 4 months. In LaSE, each fellow had to work with their employing organisation to arrange their placements both in GP and urgent care, resulting in a variety of lengths of placement and experience. This change meant that each fellow had more individualised programme as shown in table 2. It's worked really well for me … sorting things out myself and not just kind of fitting into a programme that exists. (Fellow 11) Open access Making sure that there's a bit of flexibility in it means that, particularly for the candidate, they will get the best experience rather than just having a rigid 'you will do this, you will do that'. (Host 02) Most fellows viewed this adaptation positively, but some were left without the anticipated spectrum of exposure and experience; for example, some fellows were placed in one service, such as an emergency department, for the year without opportunity to rotate around other services. There was a balance to be made between flexibility and creating the variety of opportunities for experience that were expected. I think the one thing, speaking to my other colleagues, is that there seems to be such variability in how the posts are in the fellowship … other fellows get to rotate a bit more and I think I would have liked to have rotated into other posts as well. (Fellow 13) If you make it too rigid then you deny them the opportunity of opportunistic learning but if you make it too fuzzy then everybody has a very individual experience. (HEE09) There were mixed feelings about the length of placements, but it was generally felt longer placements enabled better embeddedness and in-depth learning, particularly in general practice: I think being in one department for a whole year will perhaps give us more time to familiarise ourselves and actually produce some meaningful project work I think as well. (Fellow 14) If the GP placements could be sort of a whole year rather than six months because it sounds a bit like our fellow just kind of got going and then had to move on. (Mentor 08) Overall participants felt positive about the fellowship programme, evidenced by their willingness to consider participating in future programmes or recommending it to colleagues. Fellows reported that the programme largely met their expectations, in line with its aims (Box 1), in particular helping them with leadership skills, system understanding and upskilling them in urgent care. The positive aspects that were described were very similar to those reported for the West Midlands pilot. 17 As in the West Midlands pilot, all the fellows stated that they would recommend it to colleagues. Yes, absolutely.[recommend it to others] I think it offers good experience in terms of just more variety to the GP work and good learning from the academic point of view and working with the CCGs. (F12) Negative feedback centred on frustrations over lengthy contracting issues, relating to funding alterations, and the changes to placements discussed above. DIsCussIOn This study has shown that a 1-year urgent/emergency care fellowship programme, developed in one region to address workforce challenges facing the NHS, can be successfully transferred to other contrasting areas. Through retaining core elements of the programme but being flexible in their implementation, fellows experienced a more variable but, in the main, equally valuable experience. In so doing, the programme appears to be successfully addressing the needs expressed by many newly qualified GPs who feel underprepared in managing patients with multi morbidities, 9 and lacking expertise in management, leadership and quality improvement. [19][20][21][22] The changes to the funding model resulted in concomitant changes to the arrangements of placements, leading to benefits and challenges. The new funding model should ensure the programme's sustainability, but a consequence was that greater priority is now placed on meeting host organisations' expectations and at times this negatively affected fellows' clinical placements. Increased flexibility in placement options enabled some fellows to tailor placements to their interests; however, others reported a lack of breadth in their clinical experience or control over where they were placed. Including access to placements in commissioning bodies and through being involved in quality improvement projects, the programme gave fellows experiences that go beyond the scope of GP vocational training. While time will tell the extent to which the fellowship programme develops future leaders, participants felt that the scheme was relevant to achieving this aim in the same way as had been evidenced by the West Midlands pilot. 17 Most of the fellows at LaSE stated they would be looking for future positions encompassing clinical and leadership roles, with some from the first cohort already securing them. The broadening of the programme to include multiprofessional fellows was welcomed with all groups seeing the benefits of cross-disciplinary learning. However, more guidance is required for host organisations on professional skillsets to maximise placement opportunity and satisfaction, including the need to understand it is not intended to be a competency-based programme. Research on innovation and service change in the NHS has shown that there are many, wide ranging, factors that affect successful adoption, the complexity of which has been demonstrated. 23 Common to many studies is the need for champions who take the innovation forward while the likelihood of success is improved with more senior champions. 11 15 16 NHS organisations often rely on individuals taking on the role of champion as an additional task whereas innovation in other industries tends to be seen as a specialism it its own right. 24 The need for adequately funded innovation projects alongside investment in capacity, skills and leadership has also been found crucial to successful adoption. 25 26 The transfer of the fellowship from the West Midlands to LaSE benefited from key senior champions within HEE who drove the project forward. Where there were issues Open access in securing placements, these could potentially be overcome with better understanding of the programme in secondary care and the co-opting of champions in host organisations. Another key element of successful innovation is reported to be a programme open to adaptation, refinement or modification. 23 This research showed how the fellowship programme could be adapted to suit local needs in different areas without losing its core elements. strengths and limitations The study had access to all the fellows that participated in the fellowship programme in 2016/2017 in LaSE, with 15 of the 17 fellows engaging with the evaluation. This gives strength to the representativeness of the views reported. Fellows were followed up on a number of occasions giving the opportunity to understand their experience at various stages of the fellowship. The study successfully collected views and expectations from the perspective of a wide range of individuals who worked with the fellows, giving depth to the findings. One limitation of the study was the small number of non-GP fellows which precluded the separate analysis of this group. A further limitation was the time period over which the work was undertaken as we were unable to follow-up fellows over a long period of time after their programme had ended, therefore, cannot report how they were able to apply their experience in subsequent practice. Although the study was limited to assessing the transferability of the programme from one region to another, the West Midlands and LaSE together cover 31.8% of the population of England 27 and include 5 of the 13 local areas within two of the four regional Local Education Training Boards. Hence, it is likely that the findings have relevance to the rest of the country. The financial model supporting the scheme was shown to be of fundamental importance to the success of the programme, influencing the way that clinical placements were identified and developed. However, it was beyond the scope of the study to undertake an economic evaluation of the programme. While this is an important consideration, the costs and benefits of the scheme need to be viewed over the medium to longer term in relation to how the fellowship is preparing clinicians to meet future workforce requirements, in addition to the return that fellows give to host organisations in the short term. Implications for practice There is a clear need for training for GPs and other primary care professionals in order to prepare for future NHS workforce needs. The evaluation of this fellowship programme demonstrates a model of training that is well received and accepted by fellows and those who work with or employ them. It appears to be suited to delivery within widely varying settings hence addressing the call for 250 fellowship placements to be made available across England. 4 It could be modified to provide experience in a range of other priority clinical areas, such as mental health or frailty. This study highlights how it can be successfully adapted to fit with local funding and service requirements, while maintaining the balance with academic and leadership training and general practice experience. It has also shown the benefit of widening the programme to other primary care professional groups, although identified that careful consideration needs to be given to the choice of clinical placements. Cross-sector working will be increasingly important given growing numbers of individuals with multimorbidity and complex health needs being treated in primary care, and programmes like this will be valuable in building cross-sector and interprofessional understanding. In conclusion, we have shown that a 1-year fellowship programme can be successfully transferred from one NHS region to another if flexibility and adaptation are enabled. The broader benefits that such fellowship schemes have to the participating health service organisations need further investigation.	2019-03-11T17:17:58.489Z	2019-01-01T00:00:00.000	{ "year": 2019, "sha1": "40811ab967bb8b7a48ec47640a375c520a2bb7f9", "oa_license": "CCBYNC", "oa_url": "https://bmjopen.bmj.com/content/bmjopen/9/1/e023384.full.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "6c732cad1388d87955560ca99d5d0576d0ce6bf4", "s2fieldsofstudy": [ "Education", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
254974270	pes2o/s2orc	v3-fos-license	Geometry-Based Phase and Time Synchronization for Multi-Antenna Channel Measurements Synchronization of transceiver chains is a major challenge in the practical realization of massive MIMO and especially distributed massive MIMO. While frequency synchronization is comparatively easy to achieve, estimating the carrier phase and sampling time offsets of individual transceivers is challenging. However, under the assumption of phase and time offsets that are constant over some duration and knowing the positions of several transmit and receive antennas, it is possible to estimate and compensate for these offsets even in scattering environments with multipath propagation components. The resulting phase and time calibration is a prerequisite for applying classical antenna array processing methods to massive MIMO arrays and for transferring machine learning models either between simulation and deployment or from one radio environment to another. Algorithms for phase and time offset estimation are presented and several investigations on large datasets generated by an over-the-air-synchronized channel sounder are carried out. I. INTRODUCTION Massive multiple-input multiple-output (MIMO) is widely accepted to be a crucial technology for increasing the spectral efficiency of future cellular wireless systems through spatial multiplexing. Distributed massive MIMO, where antennas are distributed along building facades or even across multiple buildings, is of particular interest thanks to the thereby greatly improved spatial macro-diversity [1]. For these deployments, frequency, time and phase synchronization of the distributed transceiver chains is an important subject. Frequency synchronization is usually straightforward to achieve using over-theair (OTA) reference signals, through the backhaul network or based on other accurate frequency references. Phase and sampling times, however, are usually only approximately synchronous over certain timeframes and only down to constant, antenna-specific offsets, which makes calibration challenging. These offsets, if they occur reciprocally for both uplink and downlink channel, are not an issue for massive MIMO operated in time division duplex (TDD) mode. Since they are constant, they are also irrelevant for most deep learningbased applications of channel data (e.g., localization), as neural networks can easily learn to compensate for them. However, phases calibrated in absolute terms are essential for classical array processing techniques (e.g., angle of arrival (AoA) estimation) and for transferring deep learning models between different environments or between simulation and reality. Phase and time offset calibration in antenna arrays is a well-known field of study with a wide range of applications, from radio and audio signal processing to seismology. For example, [2] proposes a calibration method that can estimate ultrasonic sensor array phase and gain offsets as well as mutual coupling errors based on a set of known transmitter locations in a well-controlled environment, but assumes concurrent transmission from multiple locations. In addition to estimating gain, phase and mutual coupling errors, [3] also accounts for sensor location errors within a planar area, but assumes known transmit signal phases. In [4], gain, phase and mutual coupling uncertainties are corrected in a self-calibration step without the need for known transmitter locations, but only within a single antenna array. The contribution of this paper is to solve the sensor array calibration problem under conditions that are typical for distributed channel sounder measurements: • There are multiple antenna arrays distributed in space and their location is precisely known. • Synchronization has been achieved except for a constant, antenna-specific phase and sampling time offset. • Channel measurements with unknown starting phase for multiple known transmitter locations are available. We propose to estimate local oscillator (LO) phase and sampling time offsets in the context of an orthogonal frequency division multiplex (OFDM) system with the help of the known antenna locations based on the assumption of strong lineof-sight (LoS) propagation paths. We show that geometrybased calibration is possible despite the distributed system architecture and the unknown transmitter phases. In Sec. II, we introduce our channel sounder. The calibration problem is stated in Sec. III and an estimation procedure based on one of two different algorithms is derived in Sec. IV. Sec. V presents some experimental results and finally, the performance of the two estimation algorithms at low signal-to-noise-ratios (SNRs) is compared in Sec. VI. Throughout the paper, we use the notations defined in Tab. I. II. OUR CHANNEL SOUNDER To tackle some of the challenges of distributed massive MIMO, we developed a distributed massive MIMO channel sounder called Distributed Channel Sounder by University of Stuttgart (DICHASUS), presented in [5]. DICHASUS can capture large channel state information (CSI) datasets 1 , containing frequency-domain channel transfer functions. The transfer functions are measured between multiple, potentially spatially distributed, receiver antennas and a mobile transmitter (MOBTX) in the field. Channel measurements are tagged with accurate "ground truth" data including antenna locations. Positions are obtained using a robotic total station (tachymeter) with an accuracy on the order of a few millimeters or less. DICHASUS uses OTA frequency, time and phase calibration for the receivers, based on a reference transmitter (REFTX) broadcast. REFTX and MOBTX operate continuously and are multiplexed in frequency. Thanks to real-time adjustments and a software-based postprocessing step, the antenna array receiver (AARX) chains are almost perfectly synchronous to their respective received REFTX signal in both LO and sampling clock frequency and phase. However, as illustrated in Fig. 1, this does not necessarily imply that they are synchronous in absolute terms, since the OTA synchronization broadcast passes through different, unknown channels between REFTX and each AARX antenna, leading to the aforementioned antenna-specific constant phase and time offsets. The datasets measured by our DICHASUS channel sounder contain channel coefficient estimates for each of the N sub OFDM subcarriers of the MOBTX signal measured at all L AARX antennas and at D different time instances. Furthermore, the 3-dimensional MOBTX position x d ∈ R 3 at each time instance d is recorded. The location (and orientation) of each receive antenna y ∈ R 3 , ∈ {1, . . . , L} can be reconstructed from metadata. For the following derivations, it is beneficial to split the dataset by subcarrier into matrices 1 All datasets are available at https://dichasus.inue.uni-stuttgart.de III. PROBLEM STATEMENT A. Timing Impairment Model As long as the sampling time offsets t off, for antennas ∈ {1, . . . , L} are sufficiently small, their effect in an OFDM system can be modelled simply as a linear phase rotation over the subcarrier index n. With g [n] ∈ C denoting the phase and gain offset experienced by antenna for subcarrier n, ϕ off, ∈ [0, 2π) as the phase offset of the lowermost subcarrier at n = 0, t off, as the sampling time offset of antenna and ∆f sub being the MOBTX subcarrier spacing, this linear phase shift can be expressed as In (1), z are residual error terms. The objective of our calibration is to obtain estimates for ϕ off, and t off, from R[n], n ∈ {1, . . . , N sub } and x d , d ∈ {1, . . . , D}. B. Antenna Model Our calibration method is based on comparing measured channels to ideal line-of-sight channels computed based on transmit and receive antenna locations x d and y . It is therefore dependent on the antenna type and the antenna model. The measurements analyzed in this work were captured using probe-fed microstrip patch antennas, shown in Fig. 2, as receive antennas and dipole antennas at the transmitters. For simplicity, we model all antennas as isotropic radiators, with the phase center at the center of the patch or dipole. This idealized model introduces several errors, most notably: • The radiation pattern of the antenna is neglected, leading to unrealistic expected amplitudes. • The true phase center of the antenna might actually be at a different location (e.g., above the patch surface). • The equiphase surface of the antenna, i.e., the set of all points at which the phase of a radiated wave is equal at a given time instance, may not be a perfect sphere. The latter two errors, illustrated in Fig. 2, are especially problematic for our calibration approach, which is based A. Computing Ideal Channel Coefficient Matrices A[n] Neglecting constant factors, the expected channel coefficient for antenna , time instance d and subcarrier index n under the assumption of a free space propagation model is given by Note that the wavelength λ[n] is subcarrier-specific, leading to a subcarrier-dependent phase rotation, which is due to the propagation delay. The ideal channel coefficients computed according to (2) are collected in the matrix B. Estimation of Phase and Gain Error Vectors g[n] The phase and gain errors of the receiver antennas for one particular subcarrier n are modelled using the vectors g[n] = (g 1 [n], . . . , g L [n]) T ∈ C L . This is in contrast to [2], which uses a matrix G, where off-diagonal entries represent mutual coupling between antennas. Here, we assume that mutual coupling is negligible and therefore only consider gain and phase offsets. The objective of this step is to estimate g[n] from A[n] and R[n]. To simplify the notation, we will omit the subcarrer index n in the following paragraphs. Calibration is performed by comparing the ideal channel coefficient vector A :d for each time instance d to the actually received vector R :d . Assuming strong LoS channels between MOBTX and AARX, the received vectors after calibration should approximately correspond to the ideal vectors except for a random phase rotation due to the transmitter's unknown starting phase, modelled as s d = e jϕ d ∈ C. The normalization \|s d \| = 1 is justified by the approximately constant transmit power throughout all time instances. With enough measurements under LoS conditions, it should be possible to perform accurate phase offset estimation, since, in a sufficiently rich scattering environment, datapoints without LoS paths effectively exhibit random phase and gain values. With s = (s 1 , . . . , s D ) T , this intuition is mathematically formalized as where Z ∈ C L×D is a matrix of additive residual errors with unknown distribution. In the following, two least squares approaches are proposed to find g based on (3). The first approach in Sec. IV-B1 directly transforms (3) into an optimization problem that is solved by coordinate descent. The second approach in Sec. IV-B2, on the other hand, does not estimate g directly, but first estimates gg H , at which point estimating g can be expressed as an eigenvalue problem. 1) "Coordinate Descent"-Based Least Squares Optimization: 2 By minimizing the sum of squared absolute residuals Z 2 F w.r.t. all unknowns, we obtain an optimization problem: min g,s In the general case, the quartic optimization problem in (4) is not convex. However, we found that for measured datasets, our algorithm always converges to the same result despite random initialization (except for global phase rotations). We propose to solve (4) using a coordinate descent approach. That is, the objective function is minimized w.r.t. just one coordinate, either g ( = 1, . . . , L) or s d (d = 1, . . . , D), at a time while all other coordinates are fixed. Here, coordinate descent leads to closed-form equations for coordinate blocks g and s, within which all coordinates can be updated simultaneously, making it particularly computationally efficient. (i) Optimize w.r.t. g i with all other coordinates fixed Since \|g a d s d − r d \| 2 is constant w.r.t. g i for = i: This is a linear least squares problem with solution: All g can be optimized simultaneously as a block, since updating g i does not impact the update for g with = i. 2) Autocorrelation and Principal Eigenvector-Based Estimation: The coordinate descent approach in Sec. IV-B1 needs several iterations to approximate the optimum when estimating g directly. An alternative, less computationally complex approach estimates the autocorrelation matrix gg H and, from it, obtains an estimate for g. Solving (3) for gg H yields with residual noise matricesZ d with unknown distribution. An estimateĈ for gg H can be obtained simply by computing the sample mean over all time instances d: Note thatĈ is Hermitian, i.e.,Ĉ H =Ĉ and, being a sum of positive semidefinite matrices, also positive semidefinite. To obtain an estimateĝ for g, we apply the least squares method: SinceĈ is Hermitian, ϑ HĈ ϑ + (ϑ HĈH ϑ) = 2ϑ HĈ ϑ ∈ R, and knowing ϑϑ H 2 F = ϑ 4 , we get: By setting the derivative w.r.t. ϑ to 0, we obtain which is a necessary condition for an optimum. Clearly, (7) is fulfilled by all vectors ϑ that are appropriately scaled eigenvectors ofĈ. Namely, they need to be scaled according to their corresponding eigenvalue λ ≥ 0 (sinceĈ is positive semidefinite), such that ϑ = √ λ. From (6) we see that the objective function is maximized if the eigenvector ϑ with the largest corresponding eigenvalue λ is chosen, i.e., the principal eigenvector. Hence, the estimateĝ is chosen out of the set {(ϑ , λ )} of eigenvectors with corresponding eigenvalues of C and scaled accordingly: Knowing g[n] ∈ C L for all subcarriers n, suitable values for ϕ off, and t off, can be estimated based on (1). We use Kay's single frequency estimator [6], which is robust against noisy phase values including phase jumps by 2π, to obtain an estimatet off, and then compute an estimate for ϕ off, usinĝ e −j2πnt off, ∆f sub . V. EXPERIMENTS ON MEASURED DATASETS The following analyses will be carried out on two datasets, both of which were captured at a carrier frequency of 1.272 GHz and with a bandwidth of 50 MHz: • dichasus-015x: Indoor office environment datset with L = 32 antennas in a single 4 × 8 antenna array [7]. • dichasus-cf0x: Indoor factory environment dataset with L = 32 antennas arranged in four 2 × 4 antenna arrays [8]. First, we investigate the behavior of the iterative coordinate descent algorithm presented in Sec. IV-B1 and compare it to the eigenvector-based solution from Sec. IV-B2. We perform this analysis on one particular subcarrier and on a subset of dichasus-cf0x called dichasus-cf02. The results are nevertheless typical for all datasets and subcarriers. We run the coordinate descent algorithm 40 times, each time with a different initialization g (0) . We compute the residual error Z (m) after the m th iteration step based on (3) and visualize the distribution of its Frobenius norm using a boxplot in Fig. 4. For comparison, we also draw the residual error achieved by the eigenvector-based estimation method (Sec. IV-B2) and the value of R F , which can be interpreted as the "worstcase" residual error that would occur if no calibration was possible. Clearly, the residual error almost always converges to a minimum after a few iterations. The result of the eigenvectorbased estimation is only slightly worse than the solution found through coordinate descent. Second, we visually demonstrate that our calibration makes the phase measurements usable for array processing. We pick a datapoint from dichasus-0152, which is a subset of dichasus-015x, where the transmitter is 3 m in front of, 1.2 m below and 1.2 m to the right of the antenna array's center and visualize the received phases over the physical positions of the antennas in the 4 × 8 array. Before calibration, no pattern is visible in the observed antenna phases (Fig. 5a). After calibration, the phases decrease from bottom right to top left (Fig. 5b), as one would expect for a phase-synchronous array. Third, we use dichasus-cf0x to visualize that locations with LoS paths to spatially distributed antenna arrays are crucial for calibration. We assign letters A to D to the arrays according to Fig. 6 and denote the set of antenna indices belonging to array B and C by B and C, respectively. We define the metric which can be intuitively interpreted in the following way: s C,d ∈ C and s B,d ∈ C are the transmitter starting phases for time instance d estimated according to (5), but only based on the antennas that are part of array C or B, respectively. Then, the absolute value \|s C,d − s B,d \| is a measure for the agreement between these two starting phase estimates. If the calibrations g are valid and a LoS channel exists, the starting phases should match. Visualizing \|s C,d − s B,d \| for some subcarrier n yields Fig. 6, which demonstrates that locations with LoS paths to both array C and B have the largest agreement in starting phase. This, in turn, implies that these locations contributed to the calibration between arrays B and C. VI. PERFORMANCE COMPARISON OF ESTIMATION ALGORITHMS AT LOW SNRS When deciding which of the the two proposed phase offset estimation algorithms to use, one should, in addition to computational complexity, also consider the accuracy of estimates at operating points with low SNRs. One possible measure for the accuracy of the estimateĝ is the squared cosine similarity betweenĝ and the true phase and gain offset vector g, expressed in logarithmic units: This definition of P \| dB is invariant to phase rotations equally applied to all components ofĝ, as it should be. Ideally, P \| dB = 0 dB, which is achieved forĝ = g. To evaluate the performance at low SNRs on the existing datasets, we only consider a single subcarrier and artificially add white Gaussian noise to the matrix of measured channel coefficients R. We compute the true vector g by estimating phase and gain offsets without artificial noise. We assume that the estimate without artificial noise is perfectly accurate (without noise, iterative and eigenvector-based estimation generate almost identical estimates). We then computeĝ using both Alg. 1 (iterative, with M = 40 iterations) and Alg. 2 (eigenvector-based) with artificial noise powers corresponding to SNRs between −12 dB and +5 dB and compute the respective cosine similarities P \| dB . The result is shown in Fig. 7: The solid lines correspond to the mean value of P \| dB , computed over 5000 different noise realizations for each SNR, and the shaded areas indicate the range from first to third quartile (across random noise realizations). It can be observed that the iterative algorithm outperforms the eigenvector-based algorithm at low SNRs. This result holds for both dichasus-cf0x (Fig. 7a) and dichasus-015x (Fig. 7b), although the performance gap in dichasus-cf0x is more pronounced, possibly due to the higher number of datapoints. While this may not always be the case, it seems that the iterative algorithm is better for most scenarios in terms of estimation accuracy. VII. CONCLUSION AND OUTLOOK A geometry-based calibration procedure was developed and experimentally shown to be applicable to distributed massive MIMO channel sounder measurements. The accuracy of two different calibration algorithms was evaluated for low-SNR operating regimes. If additional information is available, e.g., if receiver gain imbalances are known to be negligible, additional constraints could enhance the calibration's performance.	2022-06-14T06:41:09.538Z	2022-06-13T00:00:00.000	{ "year": 2022, "sha1": "993f070c19fc5c0444efb9d4c865ee8a89e4d8e6", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "993f070c19fc5c0444efb9d4c865ee8a89e4d8e6", "s2fieldsofstudy": [ "Engineering", "Physics" ], "extfieldsofstudy": [ "Computer Science", "Engineering", "Mathematics" ] }
237659958	pes2o/s2orc	v3-fos-license	Neural Network Optimization Algorithms to Predict Wind Turbine Blade Fatigue Life under Variable Hygrothermal Conditions : Moisture and temperature are the most important environmental factors that affect the degradation of wind turbine blades, and their inﬂuence must be considered in the design process. They will ﬁrst affect the resin matrix and then, possibly, the interface with the ﬁbers. This work is the ﬁrst to use a series of metaheuristic approaches to analyze the most recent experimental results database and to identify which resins are the most robust to moisture/temperature in terms of fatigue life. Four types of resin are compared, representing the most common types used for wind turbine blades manufacturing. Thermoset polymer resins, including polyesters and vinyl esters, were machined as coupons and tested for the fatigue in air temperatures of 20 ◦ C and 50 ◦ C under “dry” and “wet” conditions. The experimental fatigue data available from Sandia National Laboratories (SNL) for wind turbine-related materials have been used to build, train, and validate an artiﬁcial neural network (ANN) to predict fatigue life under different environmental conditions. The performances of three algorithms (Backpropagation BP, Particle Swarm Optimization PSO, and Cuckoo Search CS) are compared for adjusting the synaptic weights of the ANN and evaluating the efﬁciency in predicting the fatigue life of the materials studied, under the conditions mentioned above. For accuracy evaluation, the mean square error (MSE) is used as an objective function to be optimized by the three algorithms. regardless of the hygrothermal effect. All materials showed a signiﬁcant decrease at 50 ◦ C wet. Introduction Blades are one of the most critical components of wind turbines. They capture wind energy and convert it into mechanical energy for the production of electricity. Defective blades significantly affect the energy conversion efficiency of the wind turbines, and blade failures have a significant impact on the cost of energy (repair, maintenance, etc.). Therefore, the increased reliability and lifetime of wind turbine blades are important for the cost of energy reduction. Approximately 95% of the modern wind turbine blades are made of fiber-reinforced composites because of their good mechanical characteristics: high stiffness, low density, and long fatigue life [1]. Compared to alternative materials, fiber-reinforced composites have other advantages in terms of weight, cost, quality, technical feasibility, market expectation, environmental impact, and health and safety. Several key properties are dictated by the matrix resin, including fatigue strength, which is a dominant failure mode in composite material structures, leading to the breakdown of structural integrity in areas such as the trailing edge, spars, and root connections [2,3]. Hygrothermal effects on composite materials should be considered in the early phases of design; otherwise, design iterations and failures will result in a waste of time, energy, Eng 2021, 2 279 and money. Usually, the degree of sensitivity of composites to individual environmental factors is quite different. Temperature and moisture are the most significant variables to consider for the design of wind turbine blades. The primary environmental effects are on the matrix material and then, possibly, on the interface with the fibers. The purpose of this paper is to identify resins that have good resistance to temperature/moisture while providing better fatigue life. The resins studied are thermosetting polymers, including unsaturated polyesters and vinyl esters. They are both suitable for manufacturing wind turbine blades in terms of cost and low viscosity for ductile processing by resin transfer molding (RTM) [2]. This study also proposes and compares three algorithms, namely Backpropagation (BP) Levenberg Marquardt, Particle Swarm Optimization (PSO), and Cuckoo Search (CS), in combination with the popular feedforward neural network (FNN) for the prediction of fatigue life of wind turbine blades. These new combinations were used in the training of SNL/MSU database [4], developed by SNL in collaboration with Montana State University (MSU). The dataset we used is collected from their large public database build for comparative analysis of fatigue life for five resin systems. All the materials studied are used for wind turbine blades by most wind turbine manufacturers, owners, and contractors. The glass-fiber-reinforced plastic (GFRP) from E-glass fibers and thermoset polymers were the most appropriate choices for our analysis [5]. This database is the latest of a series of tests that the SNL has been publishing over the years. The novelty of our work is that it applies these combinations (BPNN, PSO-ANN, and CSNN) on this newly published data and creates a model that is capable of predicting the fatigue life of the different resin systems studied. Indeed, this study is based on a portion of the data used with the BP algorithm by [6], in order to improve the results already found. In contrast, this new model could be used in the future as an alternative to the costly lab tests, as well as to validate new experimental results. Following this introductory section, the rest of the paper is organized as follows: Section 2 explains the hygrothermal effects on wind turbine blade composite; Section 3 presents the experimental fatigue data conducted with wind turbine blades related materials; Section 4 gives a brief overview of the BP, PSO, and CS and their implementation as a training algorithm of the NN; Section 5 describes how the CSNN accurately predicts the fatigue life of wind turbine blade materials; finally, some conclusions are drawn in Section 6. Hygro-Thermo-Mechanics of Wind Turbine Blades Offshore and onshore wind turbines are exposed, depending on the climatology of the region, to climatic constraints such as variations in temperature, moisture, and sand grains associated with storms. These natural phenomena can easily damage the skin of the blade (gelcoat) [7]. However, a composite wind turbine blade is highly durable if the layer of gelcoat that protects it from the external environment has good physicochemical characteristics. The state of the problem illustrated by Figure 1 is of the hygro-thermomechanical type, because the structure of the blade is subjected to three types of loadings of origin [8]: • Mechanical (wind gust, storms...); • Thermal (temperature variation); • Hygrometric (moisture variation). where ∆ and ∆ , respectively, represent the variations of temperature and moisture. During the design process, the behavior of the gelcoat to climatic phenomena, as well as hygro-thermomechanical stresses, must be considered in order to predict the undesirable effects in the short-and/or long-term [7,8]. Figure 1. Blade subjected to a hygro-thermomechanical loading [7]. Once the plates are fabricated and nailed, the specimens can be cut. The geometry of the specimen can strongly affect the experimental results, and it is important to decide which one to use for testing. All samples were prepared using RTM. The fabrics were cut by a rolling knife and stacked in the mold following the stacking sequence "lay-up" given in each case. All specimens were machined from the plates using a water-cooled diamond saw, and their edges were sanded before conditioning. Dry specimens were stored in ambient laboratory air at an ambient temperature of 23 °C with low humidity. Other dry specimens were stored in an oven at 50 °C, and they are defined as "50 °C dry". Additionally, the wet specimens were stored in a plastic container of distilled water at 50 °C; they are defined as "50 °C wet" [2,6]. Data and Method During its operation, the blade is subject to stress variation from one cycle to another. This variation can result in a degradation of the structural resistance through the phenomena of accumulation of stresses and fatigue. A number of specimens (sections) must be manufactured in accordance with the blade structure itself. These specimens are tested under sufficient alternating load levels until rupture. A testbed and strain gauges for performing the tests are required. The experimental results obtained by Sandia National Laboratories [4] make it possible to estimate the life of the blade with a sufficiently acceptable degree of confidence. On average, this lifetime is estimated at 20 years [7]. Fatigue Data The purpose of the neural networks and optimization algorithms used in this paper is to predict the fatigue behavior of wind turbine blade composites under the hygrothermal effect and in extreme conditions (dry/wet). The specimens used for this process were tested in air temperatures of 20 °C and 50 °C. The experiments have been carried out by SNL with wind turbine-related materials and released on their website [4]. In addition, a wide variety of prospective blade materials were included in this work, including E-glassfiber/polyester and vinyl ester resins in the form of a multidirectional laminate constructions [0/ 45/0] , fiber contents (35-36%), and many stitched fabrics. Table 1 lists the types and sources of resins and reinforcement used during the manufacturing process. All materials were treated in closed molds with the RTM process, Figure 1. Blade subjected to a hygro-thermomechanical loading [7]. The constitutive equations governing the hygro-thermomechanical behavior of a stratified composite (upper and lower surfaces) and without taking into account the transverse shear are expressed according to [7] by the following compact matrix form: where ∆T and ∆m, respectively, represent the variations of temperature and moisture. During the design process, the behavior of the gelcoat to climatic phenomena, as well as hygro-thermomechanical stresses, must be considered in order to predict the undesirable effects in the short-and/or long-term [7,8]. Once the plates are fabricated and nailed, the specimens can be cut. The geometry of the specimen can strongly affect the experimental results, and it is important to decide which one to use for testing. All samples were prepared using RTM. The fabrics were cut by a rolling knife and stacked in the mold following the stacking sequence "lay-up" given in each case. All specimens were machined from the plates using a water-cooled diamond saw, and their edges were sanded before conditioning. Dry specimens were stored in ambient laboratory air at an ambient temperature of 23 • C with low humidity. Other dry specimens were stored in an oven at 50 • C, and they are defined as "50 • C dry". Additionally, the wet specimens were stored in a plastic container of distilled water at 50 • C; they are defined as "50 • C wet" [2,6]. Data and Method During its operation, the blade is subject to stress variation from one cycle to another. This variation can result in a degradation of the structural resistance through the phenomena of accumulation of stresses and fatigue. A number of specimens (sections) must be manufactured in accordance with the blade structure itself. These specimens are tested under sufficient alternating load levels until rupture. A testbed and strain gauges for performing the tests are required. The experimental results obtained by Sandia National Laboratories [4] make it possible to estimate the life of the blade with a sufficiently acceptable degree of confidence. On average, this lifetime is estimated at 20 years [7]. Fatigue Data The purpose of the neural networks and optimization algorithms used in this paper is to predict the fatigue behavior of wind turbine blade composites under the hygrothermal effect and in extreme conditions (dry/wet). The specimens used for this process were tested in air temperatures of 20 • C and 50 • C. The experiments have been carried out by SNL with wind turbine-related materials and released on their website [4]. In addition, a wide variety of prospective blade materials were included in this work, including E-Eng 2021, 2 281 glass-fiber/polyester and vinyl ester resins in the form of a multidirectional laminate constructions [0/±45/0] s , fiber contents (35-36%), and many stitched fabrics. Table 1 lists the types and sources of resins and reinforcement used during the manufacturing process. All materials were treated in closed molds with the RTM process, which were molded into their final dog-bone shape without machining. The laminate nomenclature corresponds to the Sandia National Laboratories. Therefore, the fabric details given indicate the content of stitching and transverse strands or mat to which the primary strands are stitched. Laminates were processed by RTM through resin distribution layers [9,10]; a more detailed description of the fabrication process may be found in [6,11]. Backpropagation Neural Network BPNN Artificial neural networks (ANNs) have been shown to be useful for a variety of engineering applications, including the characterization of fatigue behavior. Due to their massively parallel structure, they can solve many nonlinear and multivariate problems, for which an accurate analytical solution is difficult to obtain. The popular topology of a neural network model is illustrated in Figure 2, which typically consists of one or more input layers, output layers, and hidden layers where weights are trained. Each layer comprises one or more neurons; the neurons are interconnected so that the information passes from one layer to another, from the input layer to the output layer, through the hidden layers. Various transfer functions such as sigmoidal, linear, or triangular have been used to model neuronal activity [12,13]. Eng 2021, 2, FOR PEER REVIEW 4 which were molded into their final dog-bone shape without machining. The laminate nomenclature corresponds to the Sandia National Laboratories. Therefore, the fabric details given indicate the content of stitching and transverse strands or mat to which the primary strands are stitched. Laminates were processed by RTM through resin distribution layers [9,10]; a more detailed description of the fabrication process may be found in [6,11]. Backpropagation Neural Network BPNN Artificial neural networks (ANNs) have been shown to be useful for a variety of engineering applications, including the characterization of fatigue behavior. Due to their massively parallel structure, they can solve many nonlinear and multivariate problems, for which an accurate analytical solution is difficult to obtain. The popular topology of a neural network model is illustrated in Figure 2, which typically consists of one or more input layers, output layers, and hidden layers where weights are trained. Each layer comprises one or more neurons; the neurons are interconnected so that the information passes from one layer to another, from the input layer to the output layer, through the hidden layers. Various transfer functions such as sigmoidal, linear, or triangular have been used to model neuronal activity [12,13]. In Figure 2, is the synaptic weight matrix linking the input with the hidden layer, and is the synaptic weight matrix linking the hidden layer with the output, according to Equation (2) [6,[12][13][14]: In Figure 2, w i is the synaptic weight matrix linking the input with the hidden layer, and w s is the synaptic weight matrix linking the hidden layer with the output, according to Equation (2) [6,[12][13][14]: Eng 2021, 2 282 where w i,j,k represents the weight connection strengths for node j in the (k − 1)th layer to node i in the kth layer; out j,k is the output of node i in the kth layer and θ i,k is the threshold associated with node i in the kth layer. In Equation (3), the MSE was used as an objective function to optimize, and weights were tuned to minimize their values throughout the prediction process, defined as follows: where j is the objective function, performance, or cost function; N is the number of nodes; W is the weight matrix; y is the output obtained by the neural network; y d is the desired output. In this study, two other metaheuristic algorithms (PSO and CS) were used to avoid the local optimum trap. Both were deployed to optimize MSE by adjusting the neural network weights. The details of these two proposed algorithms are described in the next sections. Particle Swarm Optimization PSO Particle swarm optimization or PSO is an intelligent optimization algorithm; it belongs to a class of optimization algorithms called metaheuristics. PSO is based on the intelligence paradigm of swarms and is inspired by the social behavior of animals, such as fish and birds. PSO is a simple yet powerful algorithm, which has been successfully applied to various fields of science and engineering. Initially, PSO was introduced by Kennedy and Eberhart [15], where they sought to model social interactions between "particles" to achieve a given goal in a common search space, each particle having a certain capacity for memorizing and processing information. The basic rule was that there should be no conductor, or even any knowledge by the particles of all information, only local knowledge. A simple model was then developed. The algorithm works by initializing a flock of birds randomly over the searching space, where every bird is called a "particle". These particles fly with a certain velocity and find the global best position after some iteration [16]. During the flight, each particle updates its velocity vector, based on its momentum and the influence of its best position (P b ) as well as the best position of its neighbors (P g ), and then computes a new position [14]. Figure 3 briefly illustrates the concept of PSO. Eng 2021, 2, FOR PEER REVIEW 5 where , , represents the weight connection strengths for node in the ( − 1)th layer to node in the th layer; , is the output of node in the th layer and , is the threshold associated with node in the th layer. In Equation (3), the MSE was used as an objective function to optimize, and weights were tuned to minimize their values throughout the prediction process, defined as follows: where is the objective function, performance, or cost function; is the number of nodes; is the weight matrix; is the output obtained by the neural network; is the desired output. In this study, two other metaheuristic algorithms (PSO and CS) were used to avoid the local optimum trap. Both were deployed to optimize MSE by adjusting the neural network weights. The details of these two proposed algorithms are described in the next sections. Particle Swarm Optimization PSO Particle swarm optimization or PSO is an intelligent optimization algorithm; it belongs to a class of optimization algorithms called metaheuristics. PSO is based on the intelligence paradigm of swarms and is inspired by the social behavior of animals, such as fish and birds. PSO is a simple yet powerful algorithm, which has been successfully applied to various fields of science and engineering. Initially, PSO was introduced by Kennedy and Eberhart [15], where they sought to model social interactions between "particles" to achieve a given goal in a common search space, each particle having a certain capacity for memorizing and processing information. The basic rule was that there should be no conductor, or even any knowledge by the particles of all information, only local knowledge. A simple model was then developed. The algorithm works by initializing a flock of birds randomly over the searching space, where every bird is called a "particle". These particles fly with a certain velocity and find the global best position after some iteration [16]. During the flight, each particle updates its velocity vector, based on its momentum and the influence of its best position ( ) as well as the best position of its neighbors ( ), and then computes a new position [14]. Figure 3 briefly illustrates the concept of PSO. In a search space of dimension , the particle of the swarm is modeled by its vector position ⃗ = ( , , … ) , and by its velocity vector ⃗ = ( , , … ) , designating by the transpose of a matrix . The quality of its position is determined by the value of the objective function at this point. This particle keeps in memory the best position by which it has already passed, which we note ⃗ = ( , , … ) . The best position In a search space of dimension d, the particle i of the swarm is modeled by its vector position and by its velocity vector The quality of its position is determined by the value of the objective function at this point. This particle keeps in memory the best position by which it has already passed, which we note → p i = (p i1 , p i2 , . . . p id ) T . The best position reached by its neighboring particles is noted → g i = (g i1 , g i2 , . . . g id ) T . At time t, the velocity vector is calculated from Equation (4): Eng 2021, 2 283 In Equation (4), v id and p id are the velocity and position of particle i (i = 1, 2, . . . n), where n is the number of particles; d = 1, 2, . . . m, and m is the number of input variables to be optimized; w is usually a constant, called the "inertia weight factor"; c 1 and c 2 are cognitive and social acceleration factors, respectively, that scale the old velocity and increase new velocity toward P best (local best result) or G best (global best result); r 1 and r 2 represent random numbers that are uniformly distributed in the interval [0,1] [8,14]. Cuckoo Search Algorithm The cuckoo search (CS) algorithm was inspired by the brood parasitism of cuckoo birds. In fact, their breeding strategy is characterized by the laying of their eggs in the nests of other species (host birds). If the host bird discovers that the eggs are not its, it will throw them away. Otherwise, its will simply abandon its nest and build another elsewhere. CS is a metaheuristic optimization algorithm based on the following rules: • Every cuckoo lays solely one egg at a time, and the eggs are exactly set in a nest (randomly selected nest); • The nest has better quality eggs, which are carried onto the next round; • The number of the nest is fixed, and the quality of the nest is static and not alterable. In CS, each egg in a nest represents a solution, and each cuckoo can lay a single egg (which represents a solution). The goal is to use the new and potentially better solution to replace a less good solution in a nest. This metaheuristic is thus based on this parasitic behavior of the species of cuckoos associated with a logic of displacement of the "Lévy flight" type, which is specific to certain birds and certain fly species [17]. The Lévy flight is a random process in which a sequence of random steps is performed [17]. Two sequential procedures must be completed to produce random numbers with Lévy flights: step generation and selection of a random direction. One of the more efficient methods for doing so is to employ the so-called Mantegna algorithm, in which the step length s can be calculated as in Equation (5) [17,18]: Here, u and v are random values from centered Gaussian distributions; β is the scale parameter, and its recommended range is [1,2]. In the process of the cuckoo search algorithm, n randomly chosen nests come into being and the ith nest is set nest i = (x i1 , x i2 , . . . x id ), where d is the dimension of the problem. The fitness of each nest can be determined according to their own location information [17,18]. The nests are updated according to Equation (6): where α is the step size according to the scale of the problem. The produce ⊗ is the entry-wise multiplications. The random walk via Lévy flight is more efficient in exploring the search space as its step length is much longer in the long run [18]. Each nest has a certain probability (P a ) to be abandoned. If a nest is abandoned, a new nest will be created according to Equation (7): where r is a scale factor uniformly distributed between 0 and 1; nest t j and nest t k are other nests, randomly selected in ith generation [18]. Data Preparation Data preparation is important for ensuring the built model's accuracy and robustness. The accuracy of the collected data influences the precision, and the amount of data used to train the model affects robustness [19]. As mentioned above, the data were collected Table 1) in terms of fatigue lives and hygrothermal effect. The size of the database collected is 122 sets distributed over the four resins in question (ortho-polyester, iso-polyester, vinyl ester 411C-50 and vinyl ester 8084). Before we began to model the data, we had to normalize it to be in a range from 0 to 1. Since the number of cycles to failure ranged from 16 to 6,000,000 cycles, and the maximum applied compressive stress was −13.8 to −41.4 MPa, we used the range-normalized absolute difference method based on the Equation (8): This method was applied to both data ranges (number of cycles and max stress). Indeed, the starting point of the parameter optimization process starts with error estimation as a difference between real and predicted values. These two sets of values must, therefore, have the same range. The normalization of inputs in neural networks can, for practical reasons, make learning faster and reduce the chances of being trapped in local minima. We used Excel to process the data from Table 2, and the results are shown in Figure 4. We started sorting the data in ascending order based on hygrothermal conditions and the resin types. We then normalized them to allow the application of the proposed neural network combinations mentioned earlier. We identified the properties extracted from the prediction for each resin, such as MSE, number of epochs, etc. This Excel file ( Figure 4) synthesizes all the data and allows us to process the simulations efficiently. Data Preparation Data preparation is important for ensuring the built model's accuracy and robustness. The accuracy of the collected data influences the precision, and the amount of data used to train the model affects robustness [19]. As mentioned above, the data were collected from the extensive database of Sandia National Laboratories to compare four types of thermosetting resins (cited in Table 1) in terms of fatigue lives and hygrothermal effect. The size of the database collected is 122 sets distributed over the four resins in question (ortho-polyester, iso-polyester, vinyl ester 411C-50 and vinyl ester 8084). Before we began to model the data, we had to normalize it to be in a range from 0 to 1. Since the number of cycles to failure ranged from 16 to 6,000,000 cycles, and the maximum applied compressive stress was −13.8 to −41.4 MPa, we used the range-normalized absolute difference method based on the Equation (8): This method was applied to both data ranges (number of cycles and max stress). Indeed, the starting point of the parameter optimization process starts with error estimation as a difference between real and predicted values. These two sets of values must, therefore, have the same range. The normalization of inputs in neural networks can, for practical reasons, make learning faster and reduce the chances of being trapped in local minima. We used Excel to process the data from Table 2, and the results are shown in Figure 4. We started sorting the data in ascending order based on hygrothermal conditions and the resin types. We then normalized them to allow the application of the proposed neural network combinations mentioned earlier. We identified the properties extracted from the prediction for each resin, such as MSE, number of epochs, etc. This Excel file (Figure 4) synthesizes all the data and allows us to process the simulations efficiently. Proposed Hybrid Models for Fatigue Life Prediction In order to create a model that can predict the fatigue life of the various resins investigated with reasonable accuracy, a total number of 122 experimental fatigue data entries were used in the training process; the parameters of input and output are tabulated in Table 3. Table 3. Input and output parameters. Input Parameters Output Parameters Normalized maximum compressive stress σ max Normalized number of cycles to failure N The training data were split into three different sets to fulfill the training requirements and improve accuracy, according to the "early stopping" method, integrated by default in the Matlab software (R2016b). Accordingly, 60% of the data serve as the training set, 20% for validation, and the remaining 20% for testing purposes. The NN used to predict the fatigue life of all resin matrices is a two-layer feedforward network with one input, the normalized maximum compressive stress σ max , and one output, the normalized number of cycles to failure N. The network consists of a single hidden layer of 10 neurons using a sigmoid activation function, while the output uses a linear activation function with one computation neuron. It is a fixed architecture for all our proposed combinations, and we try each time to replace the BP algorithm (Levenberg-Marquardt) with one of the suggested algorithms (PSO and CS). Since we have only one input and one output, data scaling was achieved by normalizing/standardizing real-valued input and output variables. The effect of varying the number of hidden neurons on the prediction of fatigue life was investigated. In each test, the number of neurons in the hidden layer was changed to obtain the lowest root mean square error (RMSE). Figure 5 shows the variations of RMSE according to the number of hidden neurons for 14 trials. Eng 2021, 2, FOR PEER REVIEW 8 In order to create a model that can predict the fatigue life of the various resins investigated with reasonable accuracy, a total number of 122 experimental fatigue data entries were used in the training process; the parameters of input and output are tabulated in Table 3. Table 3. Input and output parameters. Input Parameters Output Parameters Normalized maximum compressive stress Normalized number of cycles to failure The training data were split into three different sets to fulfill the training requirements and improve accuracy, according to the "early stopping" method, integrated by default in the Matlab software (R2016b). Accordingly, 60% of the data serve as the training set, 20% for validation, and the remaining 20% for testing purposes. The NN used to predict the fatigue life of all resin matrices is a two-layer feedforward network with one input, the normalized maximum compressive stress , and one output, the normalized number of cycles to failure . The network consists of a single hidden layer of 10 neurons using a sigmoid activation function, while the output uses a linear activation function with one computation neuron. It is a fixed architecture for all our proposed combinations, and we try each time to replace the BP algorithm (Levenberg-Marquardt) with one of the suggested algorithms (PSO and CS). Since we have only one input and one output, data scaling was achieved by normalizing/standardizing real-valued input and output variables. The effect of varying the number of hidden neurons on the prediction of fatigue life was investigated. In each test, the number of neurons in the hidden layer was changed to obtain the lowest root mean square error (RMSE). Figure 5 shows the variations of RMSE according to the number of hidden neurons for 14 trials. PSO-Based ANN A neural network optimized by a particle swarm algorithm is also called a PSO-ANN (particle swarm optimization-based artificial neural network) combination. This algorithm takes the weights and biases of the trained neural network as a single particle. For training a neural network using the PSO, the fitness value of each swarm particle (member) is the value of the error function evaluated at the particle's current position, and the particle's position vector corresponds to the network's weight matrix [14,20]. The number PSO-Based ANN A neural network optimized by a particle swarm algorithm is also called a PSO-ANN (particle swarm optimization-based artificial neural network) combination. This algorithm takes the weights and biases of the trained neural network as a single particle. For training a neural network using the PSO, the fitness value of each swarm particle (member) is the value of the error function evaluated at the particle's current position, and the particle's position vector corresponds to the network's weight matrix [14,20]. The number of PSO dimensions is taken as the total number of neural network weights and biases. The PSO-ANN combination procedure is illustrated in the flowchart of Figure 6. Returning to the parameters, to improve the convergence rate and the learning process, the parameters presented in Table 4 were better suited for the execution of the PSO algorithm. In this combination, the number of dimensions of the PSO domain corresponds to the number of weights and biases of the neural network, from which each PSO dimension corresponds to a weight or bias of the neural network. Since it is not possible to display more than three dimensions, we will only show the first and the last dimension in the x and y axes. The z-axis corresponds to the performance function of the neural network, which also represents the objective function of the PSO (Figure 7). The goal of PSO will be to optimize the MSE of our network. Returning to the parameters, to improve the convergence rate and the learning process, the parameters presented in Table 4 were better suited for the execution of the PSO algorithm. In this combination, the number of dimensions of the PSO domain corresponds to the number of weights and biases of the neural network, from which each PSO dimension corresponds to a weight or bias of the neural network. Since it is not possible to display more than three dimensions, we will only show the first and the last dimension in the x and y axes. The z-axis corresponds to the performance function of the neural network, which also represents the objective function of the PSO (Figure 7). The goal of PSO will be to optimize the MSE of our network. Eng 2021, 2, FOR PEER REVIEW 10 which also represents the objective function of the PSO (Figure 7). The goal of PSO will be to optimize the MSE of our network. For our topology, we have 31 dimensions where the PSO particles search an optimum. These 31 dimensions correspond to the weights and biases of our neural network (two layers, one of 10 and the other of a single neuron, give 20 weights and 11 biases). In this search space, each particle will have projections on the dimensions that correspond to the parameters (weight and bias) of the NN. In our case, the number of parameters is 31 (20 weights and 11 biases). Therefore, if we want to follow the evolution of the position of a particle (choose a particle among the 20 used in our PSO-ANN combination), we will see that this evolution will have 31 projections. Each projection corresponds to a parameter (weight or bias) of the NN. Figure 8 represents the change of the position of particle 9 in projection on the 31 parameters (or dimensions); dim1...dim31 represent the parameters of the NN (weight and bias). Similarly, if we want to follow the value of the objective function of a particle, the projection on the z-axis will give us the value of the objective function at any iteration. This value converges when the particles are clustered near an optimum. Figure 9 shows the evolution of the objective function according to particle 9. For our topology, we have 31 dimensions where the PSO particles search an optimum. These 31 dimensions correspond to the weights and biases of our neural network (two layers, one of 10 and the other of a single neuron, give 20 weights and 11 biases). In this search space, each particle will have projections on the dimensions that correspond to the parameters (weight and bias) of the NN. In our case, the number of parameters is 31 (20 weights and 11 biases). Therefore, if we want to follow the evolution of the position of a particle (choose a particle among the 20 used in our PSO-ANN combination), we will see that this evolution will have 31 projections. Each projection corresponds to a parameter (weight or bias) of the NN. Figure 8 represents the change of the position of particle 9 in projection on the 31 parameters (or dimensions); dim1 . . . dim31 represent the parameters of the NN (weight and bias). Eng 2021, 2, FOR PEER REVIEW 10 which also represents the objective function of the PSO (Figure 7). The goal of PSO will be to optimize the MSE of our network. For our topology, we have 31 dimensions where the PSO particles search an optimum. These 31 dimensions correspond to the weights and biases of our neural network (two layers, one of 10 and the other of a single neuron, give 20 weights and 11 biases). In this search space, each particle will have projections on the dimensions that correspond to the parameters (weight and bias) of the NN. In our case, the number of parameters is 31 (20 weights and 11 biases). Therefore, if we want to follow the evolution of the position of a particle (choose a particle among the 20 used in our PSO-ANN combination), we will see that this evolution will have 31 projections. Each projection corresponds to a parameter (weight or bias) of the NN. Figure 8 represents the change of the position of particle 9 in projection on the 31 parameters (or dimensions); dim1...dim31 represent the parameters of the NN (weight and bias). Similarly, if we want to follow the value of the objective function of a particle, the projection on the z-axis will give us the value of the objective function at any iteration. This value converges when the particles are clustered near an optimum. Figure 9 shows the evolution of the objective function according to particle 9. Similarly, if we want to follow the value of the objective function of a particle, the projection on the z-axis will give us the value of the objective function at any iteration. This value converges when the particles are clustered near an optimum. Figure 9 shows the evolution of the objective function according to particle 9. 288 Eng 2021, 2, FOR PEER REVIEW 11 Figure 9. Evolution of the objective function according to particle 9. CS-Based NN (CSNN) In CSNN, CS is used to optimize the backpropagation (BP) network's initial weights and biases. More precisely, the BP network is regarded as the objective function (fitness function), and the weights and biases are calculated by the CS algorithm to maximize the objective function [21]. It is expected that these optimal weights and biases used for the BP network will be significantly higher than the default BP network. In the first epoch, CS initializes the best weights and biases and then transfers it to the BPNN. Then, BPNN weights are calculated and compared with the best solution in the backward direction. In the next cycle, CS will update the weights with the best possible solution, and CS will continue to search for the best weights until the network's last cycle/epoch is reached or the MSE is achieved [22]. The flowchart of the CSNN combination is described in Figure 10. CS performance is tested with a discovery rate = 0.15 to optimize weights and bias and a small population size of 20. For each prediction, the trial is limited to 1000 epochs, and the minimum error is kept close to 0. As intelligent algorithms always have certain randomness, different results will be generated by each run. For each case, 10 CS-Based NN (CSNN) In CSNN, CS is used to optimize the backpropagation (BP) network's initial weights and biases. More precisely, the BP network is regarded as the objective function (fitness function), and the weights and biases are calculated by the CS algorithm to maximize the objective function [21]. It is expected that these optimal weights and biases used for the BP network will be significantly higher than the default BP network. In the first epoch, CS initializes the best weights and biases and then transfers it to the BPNN. Then, BPNN weights are calculated and compared with the best solution in the backward direction. In the next cycle, CS will update the weights with the best possible solution, and CS will continue to search for the best weights until the network's last cycle/epoch is reached or the MSE is achieved [22]. The flowchart of the CSNN combination is described in Figure 10. Eng 2021, 2, FOR PEER REVIEW 11 Figure 9. Evolution of the objective function according to particle 9. CS-Based NN (CSNN) In CSNN, CS is used to optimize the backpropagation (BP) network's initial weights and biases. More precisely, the BP network is regarded as the objective function (fitness function), and the weights and biases are calculated by the CS algorithm to maximize the objective function [21]. It is expected that these optimal weights and biases used for the BP network will be significantly higher than the default BP network. In the first epoch, CS initializes the best weights and biases and then transfers it to the BPNN. Then, BPNN weights are calculated and compared with the best solution in the backward direction. In the next cycle, CS will update the weights with the best possible solution, and CS will continue to search for the best weights until the network's last cycle/epoch is reached or the MSE is achieved [22]. The flowchart of the CSNN combination is described in Figure 10. CS performance is tested with a discovery rate pa = 0.15 to optimize weights and bias and a small population size of 20. For each prediction, the trial is limited to 1000 epochs, and the minimum error is kept close to 0. As intelligent algorithms always have certain randomness, different results will be generated by each run. For each case, 10 tests are run CS performance is tested with a discovery rate pa = 0.15 to optimize weights and bias and a small population size of 20. For each prediction, the trial is limited to 1000 epochs, and the minimum error is kept close to 0. As intelligent algorithms always have certain randomness, different results will be generated by each run. For each case, 10 tests are run to obtain a standard statistical result. The results of the network will be saved for each test. Results and Discussion In this section, we have used neural networks combined with three different algorithms (BP, PSO, and CS), called BPNN, PSO-ANN, and CSNN, respectively, as described in Section 4. These three hybrid models have been exploited to evaluate their effectiveness in predicting the fatigue life of wind turbine blade materials. The materials in question are composites based on thermosetting resins, namely the ortho-polyester resin (CoRezyn 63-AX-051), the iso-polyester resin (CoRezyn 75-AQ-010), and the two vinyl ester resins (Derakane 411C-50 and Derakane 8084). In the first step, we have trained our NN with the BP algorithm with three different temperatures: 50 • C wet, 20 • C dry, and 20 • C wet, in order to preselect the best architecture for our network, as it will follow us throughout the evaluation and prediction process. The choice of a suitable architecture ends when the network converges or reaches a global minimum already predefined as an objective function (MSE). For the materials considered, both experimental and predicted results obtained with BPNN, PSO-ANN and CSNN are compared in Figures 11-14 and show the relationship between the maximum applied compressive stress (σ max ) and the number of cycles to failure (N). For illustration and comparative purposes, we have presented in the same figure and for each studied temperature, four plots for different experimental and predicted values obtained with BPNN, PSO-ANN, and CSNN, where they show typical fatigue life predictions. We did not have enough data to predict scenarios for 50 • C wet conditions. All the data originates from SNL, who performed the experiments and made their database publicly available. Therefore, we have not being able to test the 50 • C wet with the following materials: Derakane 8084, CoRezyn 75-AQ-010 and Derakane 411C-50. Eng 2021, 2, FOR PEER REVIEW 12 tests are run to obtain a standard statistical result. The results of the network will be saved for each test. Results and Discussion In this section, we have used neural networks combined with three different algorithms (BP, PSO, and CS), called BPNN, PSO-ANN, and CSNN, respectively, as described in Section 4. These three hybrid models have been exploited to evaluate their effectiveness in predicting the fatigue life of wind turbine blade materials. The materials in question are composites based on thermosetting resins, namely the ortho-polyester resin (CoRezyn 63-AX-051), the iso-polyester resin (CoRezyn 75-AQ-010), and the two vinyl ester resins (Derakane 411C-50 and Derakane 8084). In the first step, we have trained our NN with the BP algorithm with three different temperatures: 50 °C wet, 20 °C dry, and 20 °C wet, in order to preselect the best architecture for our network, as it will follow us throughout the evaluation and prediction process. The choice of a suitable architecture ends when the network converges or reaches a global minimum already predefined as an objective function (MSE). For the materials considered, both experimental and predicted results obtained with BPNN, PSO-ANN and CSNN are compared in Figures 11-14 and show the relationship between the maximum applied compressive stress ( ) and the number of cycles to failure ( ). For illustration and comparative purposes, we have presented in the same figure and for each studied temperature, four plots for different experimental and predicted values obtained with BPNN, PSO-ANN, and CSNN, where they show typical fatigue life predictions. We did not have enough data to predict scenarios for 50 °C wet conditions. All the data originates from SNL, who performed the experiments and made their database publicly available. Therefore, we have not being able to test the 50 °C wet with the following materials: Derakane 8084, CoRezyn 75-AQ-010 and Derakane 411C-50. In all these figures, it can be noted that the prediction plots follow their experim reference, which demonstrates the efficiency of the combinations we propose. How the CSNN plot is almost superimposed on that of the experimental, because of its si ity in using mathematical models in comparison with PSO-ANN, as well as its ab adjust weights without determining a gradient by comparing with BPNN. In the tr In all these figures, it can be noted that the prediction plots follow their experimental reference, which demonstrates the efficiency of the combinations we propose. However, the CSNN plot is almost superimposed on that of the experimental, because of its simplicity in using mathematical models in comparison with PSO-ANN, as well as its ability to adjust weights without determining a gradient by comparing with BPNN. In the training process, we found that BPNN and CSNN converged more quickly to a minimal error, whereas PSO-ANN requires a considerable number of iterations to do so. In fact, this paper is the result of the sum of our experiences, which allowed us to overcome the shortcomings of our previous work [6,14,23]. In terms of error, we further evaluated the erroneously predicted values for the three combinations and presented their prediction errors in Table 5 gives better results than PSO-ANN due to its simplicity in using mathematical models. Additionally, CSNN adjusts the weights without determining a gradient, as is the case with BPNN. We found that CSNN converged faster to a minimum error in the training process, whereas PSO-ANN requires a considerable number of iterations. To make these results more precise, we had to use a bar chart to quantify the comparison. The different materials studied were compared in Figure 15 in terms of the maximum compressive stress and according to the hygrothermal conditions. The values shown in the figure were taken directly from the source database without normalization. According to the figure, the iso-polyester and the vinyl ester 411 C-50 have a high fatigue strength compared to the two remaining. The decrease in fatigue strength is very significant for the ortho-polyester at 20 • C. On the other hand, the vinyl ester 8084 still retains the same fatigue strength, regardless of the hygrothermal effect. All materials showed a significant decrease at 50 • C wet. parison. The different materials studied were compared in Figure 15 in terms of the maximum compressive stress and according to the hygrothermal conditions. The values shown in the figure were taken directly from the source database without normalization. According to the figure, the iso-polyester and the vinyl ester 411 C-50 have a high fatigue strength compared to the two remaining. The decrease in fatigue strength is very significant for the ortho-polyester at 20 °C. On the other hand, the vinyl ester 8084 still retains the same fatigue strength, regardless of the hygrothermal effect. All materials showed a significant decrease at 50 °C wet. Conclusions The use of artificial neural networks in the evaluation of composite materials of wind turbine blades will be considered as an alternative and economical solution. This will also allow us to reduce the duration of laboratory tests and let the network predict fatigue life. Depending on the results obtained, it is possible to propose iso-polyester and vinyl ester 411 C-50 as the most appropriate resins for an eventual design of wind turbine blades, under the hydrothermal effect, and this, from a lifetime and strength standpoint. The proposed hybrid neural network models BPNN, PSO-ANN and CSNN reproduce the same nonlinear characteristics obtained in the laboratory with an acceptable error. The critical element that can decide the quality of prediction and learning is simply the database, which must be relatively consistent for the network to predict future values with minimal error. The ANN architecture (number of neurons, topology, etc.) is also considered as an important factor deciding the quality of learning more than the learning parameters. We have pointed out that the CSNN is better than the BPNN and PSO-ANN in terms of prediction because of its simplicity in using mathematical models, as well as its ability to adjust weights without determining a gradient. CSNN gave us better fatigue life predictions, faster computation, more accuracy, and above all, a high convergence rate toward the actual global minimum. More research is needed to determine the possibility of finding specific combinations that always leads to the best predictions of fatigue life, as well as improving the accuracy of ANN using other optimization algorithms in the phase of learning.	2021-09-27T20:54:33.897Z	2021-07-05T00:00:00.000	{ "year": 2021, "sha1": "ccc99ab64e33c8cb280e761f8778e6339385c622", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2673-4117/2/3/18/pdf", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "50b3539a6a27e9a949897b6aaa943f2a501f71d9", "s2fieldsofstudy": [ "Materials Science" ], "extfieldsofstudy": [ "Computer Science" ] }
244097228	pes2o/s2orc	v3-fos-license	Multi-secular and regional trends of aquatic biodiversity in European Early Modern paintings: toward an ecological and historical significance . Works of art are testimonies to past civilizations and biodiversity, and provide fundamental information for guiding current conservation programs. The success of such programs requires an understanding of the reference state of ecosystems, which is rarely known because current references are in perpetual slippage toward the acceptance of degraded states. For this reason, international organizations are regularly alerted to the fact that fish and aquatic resources are threatened, signaling a major challenge for our societies. In this article we aim to enrich the historical and ecological knowledge of aquatic resources in Western Europe (Atlantic, North Sea, and Mediterranean Sea) by analyzing the taxonomic composition of aquatic biodiversity as represented in Early Modern paintings, using the statistical tools of numerical ecology. The geographic and temporal variations of the biodiversity represented in these paintings are interpreted according to environmental and human pressures, which we differentiate between using technical and socio-cultural “sieves.” Our results highlight the natural and anthropic factors that shape the spatial and temporal variations of the aquatic species depicted. These species belong to significantly different periods and regions, with a convergence between the origin of the paintings and the biogeographic area of the species. We show an overall decrease over time of represented taxa, and particularly of continental and freshwater species. We discuss the results in the light of previous works of historical ecology, archeology, history, and biology. Finally, we discuss the relevance and potential future contributions of the method developed herein to better understand the past reference state of aquatic socio-ecosystems. INTRODUCTION Biodiversity has been depicted in art by humans for the past 35,000 years, thus constituting a major source of inspiration for mankind.Art works and historical legacy are testimonies of past biodiversity and civilizations, and provide crucial information for the orientation of current conservation and restoration programs.Indeed, the implementation of such programs requires sufficient knowledge of the reference state, or baseline, of an ecosystem.When this baseline needs to be more accurately informed, art works from the past, reviewed from a historical point of view, can provide valuable clues (Guidetti and Micheli 2011).This is especially true for aquatic socio-ecosystems that are particularly affected by the biodiversity crisis (Worm et al. 2006), and in which the preservation of resources represents a major human and societal issue (Liquete et al. 2013, Schwerdtner Máñez et al. 2014).Indeed, previous studies have shown strong evidence of the existence of shifting baseline syndrome specific to exploited marine ecosystems (Papworth et al. 2009), meaning a generational amnesia leading to the loss of the reference state (Pauly 1995).Further to this, Jackson et al. (2001) used an innovative method involving paleoecological, archaeological, and historical data to study the long-term impact of overfishing on ecological communities.Numerous marine biologists have since engaged in historical approaches applied to exploited marine ecosystems (e.g., Schwerdtner Máñez et al. 2014), usually using trophic models to validate and interpret past variations of marine biodiversity (e.g., Lotze et al. 2011).More recently, a few freshwater ecosystems have been studied to permit the reconstruction of long-term historical changes of exploited fish populations in rivers (e.g., Lenders 2017, Haidvogl 2018) and lakes (e.g., Schmidt et al. 2011).In his review of historical ecology, Szabó (2015) classified this field as a mainly ecosystemcentered approach (in contrast to human-centered approaches), aware that the overall research trend combines ecology and anthropology.The present article clearly belongs to these last categories because it treats a typical human production: art. We propose in this article to investigate artistic representations of aquatic biodiversity using a transdisciplinary approach corresponding to the main author's specialties: ecology and history.As undertaken by Begossi and Caires (2015), we will use the occurrence of artistic representations of aquatic species as temporal and geographical indicators to better inform the past composition of these species and their abundances. Aquatic biodiversity in art: an overview in Europe Aquatic biodiversity in art is an infrequent but constant motif.In their introduction to 80 examples of fish imagery in art, Moyle and Moyle (1991) summarized the depiction of fishes in this way, and their notion was corroborated by the similar approach that Charmantier (2014) applied to crustaceans.The European cave art bestiary dating from prehistoric times consists essentially of large mammals.Depictions of aquatic biodiversity are rare, and even the fishes are difficult to identify (Cleyet-Merle 1990), with the notable exception of the salmon figuring in "l'Abri du poisson" (Eyzies de Tayac), which dates back 25,000 years and shows the typical morphology of a spawning specimen.Only a few aquatic species can be dated back to the Bronze and Iron Age Scandinavian civilizations, and they are poorly recognizable (e.g., the fish of the Kivik royal grave, Sweden, 600 years BCE).With a history of earlier artistic traditions, many Mediterranean civilizations represented aquatic biodiversity in a more realistic way: the open sea fauna of the Minoans (Crete, between 2700 and Purposes of the present study The objective of the present study is to analyze the taxonomic composition of aquatic biodiversity represented in European Early Modern paintings, using the statistical tools of numerical ecology.First, we undertake a set of explanatory analyses to identify the main variations in the pictorial representation of aquatic biodiversity across space and time.Second, statistical tests and representations are used to analyze these variations more precisely, according to the biological and biogeographical characteristics of the species.Finally, we endeavor to interpret these variations in the light of historical hypotheses related to technical and socio-cultural factors that may influence the representation of species in painting.To address ecological significance, we took into account ecosystem variations in relation with the biogeography of the species and trends in population changes. Surprisingly, such a transdisciplinary study has never been done before.A search in the Web of Science revealed only one similar study on paintings (Goddeeris et al. 2002), which concerned a single artist, Frans Snyders, and which was limited to the depiction of birds seen from a single socio-cultural angle, that of gastronomy.To our knowledge, there are only two publications that carry out statistical analyses of the frequency of fish images: the first explores archival photographs of fishing competitions (McClenachan 2009), and the second examines naturalistic drawings that do not pre-date the 19th century (Fortibuoni et al. 2010). Before a given species can be represented in a painting, many different conditions must align.Large-scale spatial and temporal variations in the depiction of aquatic biodiversity may reflect changes in climate, overexploitation, and habitat modification (Fig. 1).For this reason, we have applied two main "sieves" to sort the representations of aquatic taxa in paintings.First, a technical sieve that determines the availability of the specimen: Has it been caught with a specific fishing technique, or raised by aquaculture before being transported to the place where the artist worked?This knowledge is often made possible thanks to the depiction of special conservation techniques in the paintings.Second, a socio-cultural sieve that sorts primarily according to food preferences, interrelated with fishing, aquaculture, and conservation techniques.The aesthetic choices of the artist are also related to this last category.Our approach involves four steps: the collection of a sufficient number of paintings to enable statistical analysis, the identification of species or taxa, the study of their spatial and temporal variations, and finally, the interpretation of these variations through the identification of ecosystem variations (under natural and human pressures) and socio-cultural (or historical) sieves that are involved in the selection of the depicted species.These interpretations are described in the Discussion because they are based primarily on historical assumptions rather than on the statistical analyses performed in this study. This study of paintings representing aquatic biodiversity over three centuries and across the different regions of Europe provides the opportunity to better understand the variation of the taxonomic composition of the different periods and regions through socio-cultural and technical sieves.These elements will https://www.ecologyandsociety.org/vol26/iss4/art26/constitute relevant keys for an improved understanding of the evolution of aquatic socio-ecosystems during the Early Modern period. Studied regions The European peninsula is bounded to the north and west by the Atlantic Ocean and to the south by the Mediterranean Sea.Following Longhurst (2007), this region comprises two distinct provinces: the Northeast Atlantic Continental Shelf (NECS) and the Mediterranean Sea (MEDI), which are physically separated by the Straits of Gibraltar.Abundant oceanic precipitations feed numerous inland rivers and lakes of the Atlantic coast, to the contrary of the dry climate surrounding the Mediterranean Sea.This contrasted climate and the common geological history of the Mediterranean region explain the existence of many aquatic marine or continental endemic species (Tortonese 1985, Bianchi and Morri 2000, Boudouresque 2004, Coll et al. 2010, Tierno de Figueroa et al. 2013).The Atlantic region hosts more widely distributed species, some of which have been very important for human consumption since prehistoric times, but that are not adapted to the Mediterranean environment, e.g., Atlantic salmon (Salmo salar Linnaeus 1758), Atlantic herring (Clupea harengus Linnaeus 1758), and Atlantic cod (Gadus morhua Linnaeus 1758). Period The Early Modern period in Europe, spanning the years 1500 to 1800, is included within the longer climatic cooling period known as the Little Ice Age that began at the beginning of the 14th century and ended at the end of the 19th century (Lamb 1967, Le Roy Ladurie 1993).The effects of this climate change as represented in art, and especially in painting, have been well documented by several authors (Neuberger 1970, Burroughs 1981, Neuberger and Thornes 2005). During this 300-year period, the European population doubled (De Vries 1994).This trend was particularly significant in Northern Europe, in countries such as Ireland, Belgium, or England, which are all important fishing countries.Central European countries experienced an average progression.On the contrary, Southern countries increased in population by only twothirds for many reasons, including the Thirty Years War during the first half of the 17th century, several outbreaks of plague, the emigration of Protestants to the North, and the shift of the main trade routes from the Mediterranean to the Atlantic.This period is characterized by the intensification of exchanges both within Europe and between Europe and the rest of the world, made possible by the discovery of remote ecosystems and by the development of distant fisheries that were able to return their catches to Europe thanks to very well-controlled conservation processes, e.g., salting and smoking.As a luxury commodity, the trade in fresh seafood acquired structure at the national level with the establishment of special wholesale fishmonger routes.For instance, the "Chasse-marées" vessels were able to supply Paris with fresh seafood, 100 km from the sea (Fontaine-Bayer 1993, Robert 2018).New fishing techniques, such as benthic trawling or pair trawling, became more widespread.Inland fish farming, mostly practiced by monks, was essential for food during the 40 days of Christian Lent before Easter, when no other meat or animal products were eaten.This food ban was renewed at the beginning of this period with the Council of Trent (1545Trent ( -1563)).All these factors provided many opportunities for painters seeking inspiration from subjects of aquatic biodiversity. Just as portrait artists painted from living subjects, the Renaissance witnessed the advent of still life painting in connection with "Vanitas," symbolizing the ephemeral nature of life and associated with the Protestant religious reforms, the practice of Christian Lent, and the consumption of fish.From the 16th to the 17th centuries, art, science, and techniques progressed simultaneously, carried forward by the same people.In the 18th century, the Age of Enlightenment led to the exploration of scientific approaches, which would finally lead to the separation of art and science, a rupture that goes beyond the scope of our study.At roughly the same time, new artistic movements broke away from realism, leading finally to Impressionism in the 19th century and the production of artworks where the depiction of species as such remained non-recognizable "impressions." Data sources Most of the paintings we examined were obtained by searching the web using museum databases or grouped museums databases such as "Joconde" in France (Ministère de la Culture, France, http://www2.culture.gouv.fr/documentation/joconde/fr/pres.htm).The objective was to have the most reliable information on the artist behind each painting, the location, and the period of the work.The paintings (n = 73) were selected for their realism, and a sufficiently good quality image had to be available to enable species identification.Some museums were questioned directly to ensure the traceability of the painting and to obtain quality reproductions and copyrights. Species selection, identification, and characterization The depicted species were identified by a panel of more than 10 specialists from three different institutions (see Acknowledgements).A protocol was determined by the panel in order to adopt the level of precision they deemed necessary.For each painting, every specimen or group of specimens of the same species was identified using a figure that corresponded to one line in a data frame that was filled in by each specialist to indicate presence (1)/absence (0).Finally, many of the presumed "species" identified were not determined at the species level and are identified in the text as taxon/taxa.They were all described using the appropriate taxonomic level following TAXREF nomenclature.Any representations of shells without their living organism inside, and any obviously mounted fishes such as pufferfish (Tetraodontidae), were excluded from identification because of their obvious provenance from worldwide collections (e.g., "Cabinet of Curiosities") rather than belonging to the local available aquatic biodiversity.The few algae were also excluded they are not portrayed with the same fidelity as the fauna.The size of specimens was not retained because their relative lengths are not necessarily respected in the paintings. Every taxon was characterized by its biogeographical status and habitat group using the Ocean Biodiversity Information System (OBIS) database and the French Freshwater fish guide (Keith et al. 2020) with the following nomenclature: Biogeographical Status (European; Atlantic and North Sea; Mediterranean; Introduced); https://www.ecologyandsociety.org/vol26/iss4/art26/and Habitat Group (Continental; Migratory and Pelagic Amphihaline; Sea; Sea-Benthic; Sea-Demersal; Sea-Pelagic).When these data were available, trophic level, maximum size, weight, and lifespan were determined for fishes using FishBase (https://www.fishbase.org).Fishing gear was identified following the authors' expertise for the period. Positioning of paintings in space and time We assigned each painting to a period and a region according to the artists' attribution currently in use by the museum.The life of each artist (n = 41) was documented using Bénézit (2006) and the website of the Netherlands Institute for Art History (https://rkd.nl/en/explore/artists).The date and places of birth and death were the most consistently known elements of information, alongside the major town of activity.This distribution was determined in order to have a sufficient number of paintings for a statistical analysis of spatial and temporal variations. General distribution of the taxa across the paintings For regions and periods alike, we analyzed the distribution of taxa in two steps: (i) first, main distribution trends were highlighted using multivariate analyses and then, (ii) the relationship between these trends and the biological and biogeographical characteristics of the species were analyzed using statistical tests and comparisons of proportions.In order to highlight the distribution of taxa across the paintings, we used the non-normalized principal component analysis (PCA) available in FactoMineR (Lê et al. 2008).PCA is a dimensionalityreduction method that aims to determine the principal components of a data set, based on the variance of combined variables.We used this method herein to visualize how well the principal components fit the variables of interest (region and period), and thus to identify a pattern in the representation of taxa across space and time.Non-normalized PCA was chosen in order to keep the respective variances of each variable (data are centered but not standardized).In order to test the relationship between regions of representation and the distribution of taxa according to their characteristics (Biogeographical Status, Habitat Group, Trophic Level, and Fishing Gear), we performed Pearson's chi-squared tests based on the counts of each taxon across regions.To quantify more precisely the distribution of each taxon, we compared the proportion of their presence in the paintings of each region (in %) and calculated the difference of proportion between the two regions.Thus, if the obtained value is positive, the taxon appears mainly in the Atlantic-North Sea region, whereas a negative value indicates that the taxon is mainly represented in Mediterranean paintings.We used this method to quantify each taxon according to Biogeographical Status, Habitat Group, Trophic Level, and Fishing Gear.To analyze the relationship between the distribution of taxa and the three periods (16th, 17th, and 18th centuries), we performed the same set of chi-squared tests.We then plotted the residuals of each variable for significant chi-squared tests.To analyze the variations of the representation of each taxon individually during the three periods, we then calculated the standard deviation of the count of each taxon within the three periods. Focus on selected taxa In order to identify the most representative species of the geographical and temporal distribution of the different taxa, we selected the taxa that exhibited the most variance in the previous PCA (at least 80% of the explained variance).We then performed a second non-normalized PCA using the selected taxa. General distribution of paintings in space and time The 73 paintings have been attributed to 41 artists who were active in two regions, the Atlantic Ocean-North Sea and the Mediterranean Sea (Fig. 2a) during three periods (the 16th, 17th, and 18th centuries (Fig. 2b).Further to the distribution of the artists by city within the two regions, we identified two schools of art: the Flemish School (related to the Atlantic and North Sea) and the Italian School (related to the Mediterranean).We partitioned the 73 paintings as follows (Table 1): those representing Atlantic-North Sea species (n = 44), and those depicting Mediterranean species (n = 29).They date from the 16th (n = 23), the 17th (n = 28), and the 18th (n = 22) centuries.Note that the dataset is rather unbalanced regarding the geographic origin of the paintings.However, the Pearson's chisquared test based on the count of each taxon across the three periods was not significant for Biogeographical Status (p-value = 0.643), showing that the imbalance in the paintings dataset did not affect the distribution of taxa across the three time periods. Distribution of the taxa across regions The distribution of taxa in the Atlantic-North Sea and Mediterranean paintings was seen to depend on the biogeographical origin and the natural habitat of the taxa.Indeed, Pearson's chi-squared test (based on the count of each taxon across regions according to Biogeographical Status and Habitat Group) showed a significant relationship (respectively, p-value = 9.971 x 10-16 and 1.464 x 10-08), while tests based on Generally speaking, the region in which a taxon was represented was consistent with the geographical origin of the painting, showing that painters tended to represent the taxa they could observe locally.Regarding the habitat, Mediterranean paintings showed more pelagic taxa, suggesting that these species were probably more targeted by the Mediterranean fisheries.To visualize more precisely the distribution of each taxon according to its Biogeographical Status and Habitat Group across regions, we plotted the proportion of taxa (in %) in the given region for each painting (see Fig. 3b). Distribution of the taxa across periods The abundance (in %) of each class of taxa according to the three periods is shown in Fig. 4a.All classes except Malacostraca and Bivalvia showed a decrease between the 17th and 18th centuries.Reptilia, Petromyzontidae, and Mammalia (harbor porpoises, Eurasian otters, seals and dolphins) were depicted in the 16th and/ or 17th centuries, but disappear (except Delphinidae) from the 18th century paintings.Conversely, Gastropoda (snails) and Echinoidea (sea urchins) are present only from the 18th century. The abundance (in %) of each taxon in the three periods is plotted in Fig. 4b.To identify the taxa with the highest shifts during the periods, we performed the standard deviations of abundance across the three periods, and selected the first decile of this distribution (n = 13 taxa). Identified groups In order to identify the most representative species of the geographical and temporal distribution of the different taxa, we performed a first PCA using all the taxa of the dataset (explained variance: Axis 1 = 10.82 %, Axis 2 = 7.90%).We then selected the 20 taxa with the highest explained variance of the previous PCA (80% of the explained variance) and performed a second nonnormalized PCA by using these 20 selected taxa (explained variance: Axis 1 = 20.09%, Axis 2 = 15.98%,Fig. 5a).Axis 1 was mainly explained by taxa represented in (i) Mediterranean paintings from the 18th century, and (ii) Atlantic-North Sea paintings from the 16th and 17th centuries (99% of the explained variance).This suggests that the abundance of Atlantic-North Sea taxa decreased from the 17th.Axis 2 was mainly explained by freshwater taxa (74% of the explained variance).The PCA was thus composed of three groups: one of freshwater taxa, with the exception of C. harengus (Group 1), a second group of taxa mainly represented in Atlantic-North Sea paintings from the 16th and 17th centuries (Group 2), and a third group composed of taxa mainly represented in 18th century Mediterranean paintings (Group 3).To test if the distribution of selected taxa was dependent on region and period, we performed a Pearson's chisquared test based on a presence/absence matrix of each taxon in each painting, according to the three periods and the two regions. For both the periods and the regions, the tests were significant (pvalues = 0.013 and 1.395.10 - , respectively), showing that distributions of the taxa were statistically dependent across periods and regions Main paintings The paintings belonged to three periods and two regions (as shown by the PCAs in Fig. 5b and Fig. 5c).Note that the Mediterranean paintings tend to date from the 18th century, whereas the Atlantic paintings correspond in general to the 16th and 17th centuries.Regarding the distribution of the paintings, those that contributed the most to the variance of the distribution were P15 (Frans Snyders, Fish Stall, 1618-1621), P14 (Frans Snyders, The Fishmongers, 16th c.), and P45 (Jacob Van Nieulandt, Fishmongers, 17th c.) for axis 1 (23.02% of explained variance), and P3 (Joachim Beuckelear, The Fishmongers, 16th c.), P2 (Joachim Beuckelear, The Fishmongers, 16th c.), and P47 (Jacob Van Toorenvliet, A Fish Seller, 17th c.) for axis 2 (31.01% of explained variance). DISCUSSION The paintings selected for our study belong to two regions and three periods.The Mediterranean paintings date mainly from the 18th century, whereas the Atlantic-North Sea paintings tend to date from the 16th and 17th centuries.Despite this unbalanced sample, the bias did not influence the homogeneity of taxa distribution across periods.However, taxa distribution across regions was significantly related to their Biogeographical Status and Habitat Group.This may mean that the places where taxa were painted are consistent with their natural ranges, suggesting that these works of art may provide reliable clues to inform about reference states of the past.Following the purpose of this article, we investigated the significance of aquatic biodiversity representations from ecological and historical points of view.To determine historical significance, we divided the interpretations of variations using two "sieves": a technical one (determined by fishing techniques, aquaculture and transport), and a socio-cultural one (derived from food preferences and aesthetic choices).To address ecological significance, we took into account ecosystem variations in relation with the biogeography of the species and trends in population changes. Transport and conservation We have identified three types of containers used for the transport of fish: wooden barrels, wicker baskets, and tanks.Conditioned fish was transported in wooden barrels after being salted, smoked, or dried (Hoffman 2000).Barrels had been used since the Middle Ages and were primarily dedicated to salted salmon, dried cod, and brine herring.Thanks to this method, these species were widely transported and were consumed inland despite having been caught at sea (Hoffman 2005, Barrett et al. 2008).Such a barrel is visible in Appendix 3b ("Nature morte aux poissons", Unknown artist, Musée maritime de l'île Tatihou, Conseil départemental de la Manche).The fresh and live fish were transported in tanks, as can be seen in the painting at the bottom of Fig. 1.Tanks had also been used for this purpose since the Middle Ages, and were generally dedicated to the transport of freshwater species inland (Hoffman 2005), as is visible in the painting in Fig. 1.Note that these tanks are represented in 30% of the selected paintings.Wicker baskets were also used to transport fresh fish from the coasts to inland areas, as is visible in Appendix 3a ("Still life with fish", Abraham van Beyeren 1655, RISD museum).The baskets were lined with straw to improve the conservation of the fish during transport by horse or donkey (Robert 2018). In our results, the depicted taxa whose biogeography did not converge with the region in which they were painted provide a clue for the identification of transported species.This is the case for herring, whose representation is related to freshwater species (Group 1).Because herring was widely consumed inland from the Middle Ages on (Robert 2018), we find it depicted as being transported as fresh fish in tanks or baskets (Duhamel du Monceau and La Marre 1772), and as traveling at the same time as freshwater fish. In the Mediterranean, the transport of oysters and shellfish started in the 17th century, in particular, oysters from Tuscany (Italy), which were renowned for their quality.Oyster spat was also transported from Corsica, and from Livorno (Italy) where it was cultivated for the Grand Duchy of Tuscany.The transport of shellfish and oysters developed further during the 18th century with design improvements to the "Chasse-marées" vessels (Faget 2017). Regardless of region, we observed that the representation of Malacostracans (i.e., crustaceans) increased sharply beginning in the 18th c.We believe that this phenomenon is linked to the improvement of transport conditions that facilitated the trade, and therefore the consumption, of crustaceans throughout Europe. Fishing techniques Fishing gear is an important sieve that probably influenced the abundance of depictions of specific taxa.This is particularly true for Mediterranean Sea Mullidae (red mullets) and Triglidae https://www.ecologyandsociety.org/vol26/iss4/art26/(gurnards), which were specifically targeted as benthic species using the first trawling methods that appeared in the 15th century (Faget 2017).Indeed, important changes appeared in the Mediterranean beginning in the 15th century, characterized by innovations in fishing gear such as the Sardinal drift net (designed to catch small pelagic fishes), the Tartane trawlers (adapted to high seas fishing), longlines beginning in the late 15th century, and trawls towed by two vessels ("au boeuf " fishing) from the 18th century on (De Nicolò 2012, Faget 2015, 2017).This could explain the increasing frequency of depiction of these species during the studied period.In our study we highlight a tendency toward pelagic fishing in the Mediterranean.This may be related to the development of tuna traps in the 17th c. and Sardinal drift nets in the 15th c., both of which targeted pelagic fishes.During this period, the consumption of pelagic fish was characteristic of the diet of Mediterranean peoples (Faget 2017).Finally, the increase in the depiction of oysters (classified among the bivalves in our study) during this period may be related to the development of oyster harvesting in the Mediterranean in the 17th c., when metal dredges replaced hand gathering in shallow water (Faget 2017). This period is characterized by a general improvement in both Atlantic and North Sea shipping and fishing techniques, such as bottom trawlers, drift nets, and gill nets (Pitcher and Lam 2015), leading to the development and occasional collapse of specific local fisheries.These events were particularly recorded for cod, herring, and sturgeon fisheries, and have been the subject of reviews by Poulsen (2008), Pitcher and Lam (2015), and Lenders (2017). Aquaculture Other technical sieves that deserve mention concern the introduction and culture of species.In the Mediterranean, oyster culture began in Livorno (Italy) in the 18th century but remained peripheral.The practice consisted of introducing spat into ponds in order to later harvest the adult oysters (Faget 2017).Further north, the Portuguese oyster (Crassostrea angulata, Lamarck 1819) was introduced into the North East Atlantic from Asia in the 16th century (Grade et al. 2016), where it replaced harvests of the indigenous flat oyster (Ostrea edulis, Linnaeus 1758). Oyster culture began in the 17th century but remained peripheral until the 19th c (Buestel et al. 2009).Flat oysters appear in the selected 16th century paintings.It must be noted that differentiating between the Portuguese oyster and the flat oyster, as depicted in a painting, is not always possible.For example, in "La Raie" by Pierre-Siméon Chardin (1728, Louvre Museum) the experts we consulted could not identify with certainty the species of oyster represented.Another example of early aquaculture is the case of the common carp, which was raised inland in fishponds as early as the 12th c. (Hoffman 1995), meeting the needs of fresh fish in areas far from sea. Food preference Taxa whose natural distribution is European, but whose representation occurs mainly in paintings from a specific region, may indicate a culinary preference.For instance, taxa from Group 3 include European species that are strongly represented in Mediterranean paintings: grey mullet (Mugilidae), red mullet (Mullidae), and garfish (Belone belone, Linnaeus 1761).These are emblematic species of the Mediterranean.In particular, grey mullet are directly linked to the regional cultural identity, are consistently present in brackish water, and are easy to catch (De Nicolò 2019).Seabass (Dicentrarchus labrax, Linnaeus 1758) and squid (Loligo spp.) may also be associated with Mediterranean culinary preferences. We observed a variation in the trophic levels of represented fishes: they remained low during the 16th c. then increased throughout the 17th and 18th centuries.We hypothesize that this variation is connected to the beginning of standardizations in food preferences, and to an increase in the diversity of depicted species. Religious precepts Regarding the effect of religious precepts on fish consumption, a major event was the Council of Trent (1545-1563) in response to the Protestant Reform, for it reaffirmed the dietary rules of the Church: meat consumption was forbidden for roughly 130 days (35%) of the year, during which time only fish could be eaten.This led to an increase in the consumption of freshwater species such as pike (Esox lucius), common carp, salmon, and sturgeon (Acipenser sturio, Linnaeus 1758).Although these fish were already widely consumed in the Middle Ages (Hoffman 2005), this particular increase occurred in the period corresponding to the beginning of our study. Aesthetics The first paintings in our study date from the 16th c.This period corresponds to the European Renaissance, a movement that began in Italy in the 14th century and then spread throughout European art, especially through the Mannerist style.The Italian Renaissance masters greatly influenced the painters of Germany and the Netherlands during the 16th c. (Gombrich et al. 1997).The 17th c. was characterized by Baroque art in Italy, which also influenced Flemish artists who adopted a more demonstrative Baroque style.During the same period, Dutch painters developed the portrait and the landscape, with the objective of faithfully reproducing nature (in particular through still life works of art).We found a predominance of paintings of still lifes, and particularly of fish stalls, in our study.This kind of painting generally depicts an activity, which may be a practice, or even a profession (Blanchard 1981).Here, the activities highlighted are fishing, the preservation and transport of fish, the sale of fishery products at market, and their consumption.This suggests that the species represented are all intended to be eaten.The goal was not to represent the aquatic fauna in a systematic way, but rather to highlight the fish as food, surrounded by all the related activities and protagonists.These paintings therefore provide direct information about these practices.Concerning the other living creatures that figure in the paintings, human beings are predominant (37% of the selected paintings) and are often portrayed as fish sellers or customers.Cats and dogs are also depicted alongside the humans, either through their predatory instinct (cats stealing food) or because they are privileged companions (dogs lying close to humans, or chasing a thieving cat). A question that arises concerns other species such as whales and other large marine mammals, which are known to have been consumed, but do not appear in the paintings.These species were widely consumed in Europe from the 13th c. (Brito et al. 2019), and yet are not represented in any known painting.Our hypothesis https://www.ecologyandsociety.org/vol26/iss4/art26/ is linked to the large size of these animals, which, because of the constraints of the still life genre, makes them poorly adapted as subjects. Furthermore, we propose that certain species were depicted in the compositions not only because they were to be consumed, but also for their aesthetic qualities.For instance, salmon is often represented sliced, yet was surely not the only species that was transformed before being sold.We suggest that the orange color of the salmon flesh would have appealed to the painters as a means of bringing color to their compositions, and this regardless of the palatability of the fish.For example, a cock salmon (a male with breeding colors and a long, hooked jaw) is represented in a painting by Frans Snyders (Fig. 1) despite these fish rarely being consumed because of their inferior taste.Its presence in the painting may therefore be explained by the artist's desire to paint a fish with bright colors and an unusual shape.Indeed, painters did not always present market stalls with realism, but often constructed assemblies of different species, which were observed and studied separately by the artist.They sometimes added exotic species to their compositions, perhaps because of their aesthetic attributes.This is particularly true for the "Cabinet of Curiosities."However, because we chose to remove paintings representing species coming from worldwide collections, this type of sieve does not appear in our analyses. Likewise, some species are not depicted in painting despite being widely consumed, perhaps because of a lack of artistic appreciation.For instance, although the depiction of cod decreased from the 16th to the 18th centuries, it was consumed with increasing regularity (Pitcher and Lam 2015).We suggest that the conservation method for cod transport played a role here, and that thanks to the improvement of transport conditions in the 17th c., the cod arrived to market soon after being salted, dried, and stored in piles.It would therefore have been less recognizable as cod, and therefore less represented by painters.Finally, although this is not the subject of our study, a more indepth analysis is necessary to better explain the aesthetic choices of the painters and the possible symbolic dimensions of the species represented. Climate The studied period (16th-18th c.) is well documented regarding major climate changes that influenced the geographical distribution of aquatic species.For instance, the Little Ice Age (early 14th to late 19th c.) varied the biomass of certain species fished on European coasts, in particular cod and herring (Øiestad 1994(Øiestad , Holm et al. 2019)).This phenomenon, combined with the improvement of fishing techniques and local social and political events, led to stock changes for several taxa.For instance, Ravier-Mailly and Fromentin (2003) found evidence of negative correlations between bluefin tuna (Thunnus thynnus, Linnaeus 1758) catches and temperature in the 17th c.Other evidence indicates that episodes of "sterility" (i.e., the temporary weakening of the biomass of certain species fished on the coasts) occurred in the 18th c. in the Gulf of Marseille and coincided with periods of excessively cold winters (Faget 2010).Conversely, the Little Ice Age had positive effects on specific populations such as sturgeon (Acipenser oxyrinchus, Mitchill 1815), which is a coldadapted species that migrated into the Baltic Sea, taking advantage of a weakening of the population of the more cold water-sensitive A. sturio (Tiedemann et al. 2007).Further north, cooler periods had a positive impact on herring populations in the Atlantic during the late 17th c. (Southward et al. 1988). Habitat modifications Important habitat modifications that occurred in European inland lakes and rivers also led to the decline of many fish populations (Lotze 2007).For instance, the practice of maintaining ponds for harvesting freshwater fish began to be questioned at the end of the 18th c., for these facilities reduced the area available for the cultivation of cereals, and were also considered to be unhealthy (Abad 2006).European countries therefore undertook the draining of these ponds, encouraged by the centralized monarchical states and the declining influence of the monks who had previously maintained a large number of the ponds (Morera 2011).This resulted in a decrease of northern pike (Nilsson et al. 2014) and other pond species such as common carp, perch (Perca fluviatilis), and bream (Abramis brama).Other preindustrial societal activities in and along rivers, such as wood rafting or the generation of hydropower using watermills, modified fish habitats and impacted spawning grounds, eggs, larvae, and juvenile fish survival (Haidvogl et al. 2014).The expansion of watermill technology across Europe had a great impact on salmon populations between the Early Middle Ages and Early Modern Times (16th c.; Lenders et al. 2016).In addition, the channeling of European waterways began in the 17th c. in the Netherlands as a remedy for mobility shortcomings (Brolsma et al. 2011). The pollution of waterways linked to the artisanal or industrial activities of the 18th c. may also have affected freshwater fish populations.Waterways close to cities were seen to be pestilential and dangerous, and the link between artisanal or industrial activities and insalubrity was well established (Le Roux 2011).Pre-industrial mining and metallurgy also had a great impact on lakes and rivers from the 16th to 18th centuries (Bindler et al. 2009, Haidvogl 2018). Fish populations The set of paintings reveals that the natural range of species is mainly convergent with the place they were painted.This is particularly true for the Atlantic-North Sea taxa (Group 2) that are adapted to cold water. Examples of combined effects on documented species Although the geographical distribution of species remains constant, the abundance of populations vary over the three centuries.These variations are primarily related to diverse human activities. The most striking examples are the sturgeon, herring and cod.For example, sturgeon (A.sturio) populations were threatened by climate, fishing, and anthropic habitat modification from the 12th https://www.ecologyandsociety.org/vol26/iss4/art26/c. and from Italy to the Baltic (Hoffman 2005, Tiedemann et al. 2007).The Danube River sturgeon population, in particular, decreased in the 16th century (Guti 2008).North Sea herring fisheries were shaped by a combination of political, social, economic, and environmental factors between the 16th and 18th centuries (Poulsen 2008), and the expansion in the late 17th c. of the salted barrel herring trade operated by drift net fisheries in the North Sea is thought to have led to the serial collapse of inshore herring stocks (Pitcher and Lam 2015).Thus, complaints by fishers of depleted stocks became more frequent (Thurstan et al. 2014).In addition, the environmental history of Atlantic fisheries exhibits a series of local depletions and shifts in local fish communities (Bolster 2012).The discovery of the Grand Banks off Newfoundland, Canada in the late 15th c. allowed the exploitation of huge cod populations, and the technological improvements of the following century led to a rapid expansion of catches (Pitcher and Lam 2015). Reductions in the distribution of anadromous fishes such as sturgeon and salmon were also reported in Europe from the Middle Ages to Early Modern times, probably because of the combined effects of fisheries pressure (especially for sturgeons, which are estuarine fish caught at shallow depth with gillnets) and of pollution, habitat modification, and climate (Hoffman 2005, Tiedemann et al. 2007, Guti 2008, Lenders et al. 2016).All of these factors are clearly convergent with the diminishing number of depictions of freshwater and amphihaline migratory species in the paintings in our study. CONCLUSION This study highlights both the natural and the anthropic factors that shaped the spatial and temporal variations of aquatic species prior to Modern times in Western Europe.Although it is difficult to fully disentangle what relates to historical or ecological events, significant trends have been identified.The most obvious is that the fish represented in European paintings from the 16th to 18th centuries belong to two regions and three periods, which can be statistically discriminated.All represented taxa were intended for food and thus imply links to fishing and transport techniques.However, we illustrate that there are also environmental and sociocultural factors that should be taken into account.Regarding spatial variations, our results strongly suggest that there is a convergence between the origin of the paintings and the biogeographic area of the species that are represented.This confirms the objective of the study and the validity of the method used to inform the evolution of aquatic socio-ecosystems. Concerning temporal variations, we found an overall decrease of represented taxa and particularly of continental and freshwater species.Thanks to previous work showing that human impacts on freshwater ecosystems had already begun in the Middle Ages, we conclude that tendencies observed in our study are the result of these earlier changes.This demonstrates the importance of including marine and freshwater species (consumed both on the coast and inland), rather than using a segmenting approach to marine and freshwater environments.On the other hand, we also observed an increase in representations of pelagic species, Malacostraca and Bivalvia, which is probably more closely linked to developments in fishing and transport. These observations are corroborated by the works cited above, to which we have added a series of hypotheses concerning the effects of mixing environmental variations, technical sieves, and sociocultural sieves.It is not possible to precisely quantify the effect of any one of these factors on the depicted species, however they do open up interesting fields of investigation.For instance, to our knowledge, no meta-analysis of archaeological data on the Mediterranean aquatic species has been undertaken.There is thus a real need to integrate archeology into Mediterranean historical ecology studies.In terms of aesthetic interpretation, it is necessary to investigate the career and influences of each painter from an art history angle.Increasing the number of paintings studied would also provide more knowledge.Concerning taxa identification, some require more precise specification, e.g., pikes, shads, sturgeons, loligos, octopus, oysters, and urchins, in the search for disappeared, introduced, or recently detected species, e.g., the three species of pikes in France (Denys et al. 2014).In addition, there is a shift between poorly documented species and highly informed commercial species, e.g., sturgeon, herring, and cod, which needs to be rebalanced.This article constitutes an encouraging first step toward the emergence of multidisciplinary methodologies intended to better understand the past reference state of aquatic socio-ecosystems, using an integrative approach. Responses to this article can be read online at: https://www. Fig. 2 . Fig. 2. Locations and dates of birth and death of the artists included in the study.(a) Location of the cities of birth and death, and of the main activity of the artists included in the study with delineation of their zonal belonging.(b) Dates of birth and death of the artists, colored according to their region of origin. The Pearson's chi-squared tests based on the count of each taxon across the three periods showed a significant relationship with Habitat Group (p-value = 0.002) and Trophic Level (p-value = 9.427.10 - ), but not for Biogeographical Status (p-value = 0.643) and Fishing Gear (p-value = 0.046).Values of the residuals for chi-squared tests of Habitat Group and Trophic Level are shown in Fig.4c. Fig. 3 . Fig. 3. Representation in paintings of taxa by region.Differences between the proportions (in %) of paintings (located in the Mediterranean Sea and in the Atlantic-North Sea) where each selected taxon is represented.Left panel: taxa colored according to Biogeography.Right panel: taxa colored according to Habitat Group.Only the taxa with a difference of more than 5% between the Atlantic-North Sea and the Mediterranean Sea are shown. Fig. 4 . Fig. 4. Representation in paintings of taxa by period.(a) Relative abundance (in %) of class of taxa in the paintings according to the three periods: 16th, 17th, and 18th centuries.(b) Abundance (in %) of taxa in the paintings according to the three periods: 16th, 17th, and 18th centuries.Only the species with the highest variations are shown (based on the first decile of the distribution of standard deviation between the three periods).(c) Residual values of chi squared tests performed according to Habitat Group (p-value = 0.002) and Trophic Level (p-value = 9.427 x 10-7).Only the residuals of significant tests are shown. Table 1 . Number of paintings by period and region. For instance, S. salar, G. morhua, and C. harengus are exclusively present in the Atlantic Ocean. ecologyandsociety.org/issues/responses.	2021-11-14T16:27:53.303Z	2021-01-01T00:00:00.000	{ "year": 2021, "sha1": "64cc6368b3c1c4c9a7b406a39e1c96b029658026", "oa_license": "CCBY", "oa_url": "http://www.ecologyandsociety.org/vol26/iss4/art26/ES-2021-12740.pdf", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "89a3960a08bfb5b4ed25f6fbb585643219a91a15", "s2fieldsofstudy": [ "Environmental Science", "Art", "History" ], "extfieldsofstudy": [] }
248758213	pes2o/s2orc	v3-fos-license	MRI as a biomarker for breast cancer diagnosis and prognosis Breast cancer (BC) is the most frequently diagnosed female invasive cancer in Western countries and the leading cause of cancer-related death worldwide. Nowadays, tumor heterogeneity is a well-known characteristic of BC, since it includes several nosological entities characterized by diﬀerent morphologic features, clinical course and response to treatment. Thus, with the spread of molecular biology technologies and the growing knowledge of the biological processes underlying the development of BC, the importance of imaging biomarkers as non-invasive information about tissue hallmarks has progressively grown. To date, breast magnetic resonance imaging (MRI) is considered indispensable in breast imaging practice, with widely recognized indications such as BC screening in females at increased risk, locoregional staging and neoadjuvant therapy (NAT) monitoring. Moreover, breast MRI is increasingly used to assess not only the morphologic features of the pathological process but also to characterize individual phenotypes for targeted therapies, building on developments in genomics and molecular biology features. The aim of this review is to explore the role of breast multiparametric MRI in providing imaging biomarkers, leading to an improved differentiation of benign and malignant breast lesions and to a customized management of BC patients in monitoring and predicting response to treatment. Finally, we discuss how breast MRI biomarkers offer one of the most fertile ground for artificial intelligence (AI) applications. In the era of personalized medicine, with the development of omics-technologies, machine learning and big data, the role of imaging biomarkers is embracing new opportunities for BC diagnosis and treatment. INTRODUCTION In recent years, with the spread of molecular biology technologies and the increasing knowledge about the biological processes underlying cancer development, considerable interest in biomarkers has progressively grown. In 2016, the latest glossary released by the U.S. Food and Drug Administration (FDA) -National Institutes of Health (NIH) Biomarker Working Group in its Biomarkers, Endpoints, and other Tools (BEST) Resource, defined a biomarker as "a defined characteristic that is measured as an indicator of normal biological processes, pathogenic processes or responses to an exposure or intervention, including therapeutic interventions. Molecular, histologic, radiographic or physiologic characteristics are types of biomarkers". 1 Moreover, the introduction of established tumor biomarkers in the most recent edition of Tumor Node Metastasis (TNM) staging system by the American Joint Committee on Cancer for several tumor entities, including BC, illustrates the movement in progress towards precision approaches and therapies. 2 Within the framework of precision medicine, biomarkers become an important element for developing study methodology, research hypotheses and selectively applying scientific findings in cancer care. 1 Imaging findings were only recently officially recognized as biomarkers even if it is in the intrinsic nature of imaging to be applied in this sense. 3 BC was the most frequent cancer diagnosed among females in 2020 and breast MRI has been established as a non-invasive imaging modality for the detection, characterization and local staging of breast tumors with several recommendations including screening of high-risk females, pre-operative local staging and systemic therapy monitoring. [4][5][6] Contrast-enhanced MRI (CE-MRI), diffusion-weighted imaging (DWI) and magnetic resonance spectroscopy (MRS)-based imaging biomarkers have shown to be highly correlated with BC molecular subtypes and other prognostic and predictive factors. Furthermore, multiparametric MRI (Mp-MRI) approaches have been introduced to investigate associations of imaging biomarkers with histological types and subtypes, response to treatment, risk of recurrence and overall survival in BC patients. [6][7][8] ABSTRACT: Breast cancer (BC) is the most frequently diagnosed female invasive cancer in Western countries and the leading cause of cancer-related death worldwide. Nowadays, tumor heterogeneity is a well-known characteristic of BC, since it includes several nosological entities characterized by different morphologic features, clinical course and response to treatment. Thus, with the spread of molecular biology technologies and the growing knowledge of the biological processes underlying the development of BC, the importance of imaging biomarkers as non-invasive information about tissue hallmarks has progressively grown. To date, breast magnetic resonance imaging (MRI) is considered indispensable in breast imaging practice, with widely recognized indications such as BC screening in females at increased risk, locoregional staging and neoadjuvant therapy (NAT) monitoring. Moreover, breast MRI is increasingly used to assess not only the morphologic features of the pathological process but also to characterize individual phenotypes for targeted therapies, building on developments in genomics and molecular biology features. The aim of this review is to explore the role of breast multiparametric MRI in providing imaging biomarkers, leading to an improved differentiation of benign and malignant breast lesions and to a customized management of BC patients in monitoring and predicting response to treatment. Finally, we discuss how breast MRI biomarkers offer one of the most fertile ground for artificial intelligence (AI) applications. In the era of personalized medicine, with the development of omics-technologies, machine learning and big data, the role of imaging biomarkers is embracing new opportunities for BC diagnosis and treatment. BIOMARKERS CLASSIFICATION Medical imaging can be a source of biomarkers in diagnostic, predictive, prognostic and monitoring settings. 1 Breast imaging biomarkers can be divided into qualitative, ordinal and quantitative as shown in Table 1. Qualitative biomarkers are descriptive characteristics representative of the underlying pathologic condition. 1,9 The American College of Radiology Breast Imaging-Reporting and Data System (ACR BI-RADS) lexicon is the first and the best validated system of imaging descriptors in radiology. 10 Ordinal biomarkers are categories with intrinsic rankings that can be arranged in a meaningful order. 9 Breast MRI background parenchymal enhancement (BPE) with minimal, mild, moderate and marked categories is an example of ordinal biomarkers. Quantitative biomarkers are objective, measurable and reproducible parameters. 9 Anatomic structures 2D and 3D measurements are examples of quantitative biomarkers essential in diagnosis, staging and monitoring of response to treatment, as it is when applying the Response Evaluation Criteria in Solid Tumors (RECIST 1.1) criteria. 11 Additional quantitative biomarkers derived from breast Mp-MRI, are apparent diffusion coefficient (ADC) maps values of DWI and the transfer constant (K trans ) that provides a measure of capillary permeability with CE-MRI perfusion. The use of panels or scoring systems combining multiple imaging parameters, such as TNM, can perform significantly better than individual ones. 12,13 Further example is the increased diagnostic performance of Mp-MRI in BC molecular subtype prediction based on the underlying biological features. For instance, well-known imaging biomarkers of triple-negative BC (TNBC) are intralesional necrosis and peritumoral edema at T 2 weighted images, smooth margin and rim enhancement at CE-MRI [14][15][16][17] Figure 1. In recent years, researches have demonstrated that different BC phenotypes show specific imaging texture features. 18 Thus, the new perspective of breast MRI includes artificial intelligence (AI) applications. The intrinsic multiparametric nature of MRI has the greatest potential to incorporate AI applications into the so called precision medicine. The number of breast imaging biomarkers will increase in the next future, expanding the role of imaging in breast care. Table 2 shows different biomarkers classified according to imaging modality. CE-MRI Over the past two decades, CE-MRI has improved breast MRI diagnostic accuracy with sensitivity up to 99% and variable specificities ranging from 47 to 97% in the detection and characterization of breast lesions, 9,19 assessing breast tissue vascular microenvironment and tissue permeability. As angiogenesis plays an important role in tumor biology, CE-MRI biomarkers and pharmacokinetic parameters were widely investigated. 20 BPE is described as the enhancement of fibroglandular tissue in the dynamic CE-MRI early phases. 21,22 In the last decade, BPE has generated interest and has been added in the last edition of BI-RADS MRI lexicon that provides standard descriptors for BPE level and distribution. 10,23 It has been shown that BPE is a hormonally sensitive feature that declines over time with the onset of menopause, after oophorectomy and in patients who have been treated with tamoxifen or aromatase inhibitors. 24,25 Initial results from two case-control studies on high-risk subjects have attested that BPE can be a predictive biomarker of BC risk. In particular, in females previously stated as at high risk of BC, a marked BPE increases the personal risk of BC up to tenfold. 21,26 Kim et al found a significant correlation between qualitative BPE and epidermal growth factor receptor (EGFR)-positive BCs compared to EGFR-negative BCs. In this paper, BPE was also measured with the semi-quantitative background enhancement coefficient (BEC), evaluated through regions of interest drawn on healthy breast tissue. BEC as well as ipsilateral whole breast vascularity, were significantly higher in >2 cm tumors than in tumors smaller in size. 27 In addition, BPE may reduce breast MRI sensitivity by obscuring enhancing cancers or may decrease specificity by determining enhancement patterns that mimic the appearance of malignancies. 28,29 Nevertheless, current evidences have not confirmed a significant correlation between BPE and an increase in either false-positive or false-negative findings on breast MRI. 22,30 However, a recent systematic review highlights the wide variability in the quantitative evaluation of BPE on breast MRI, thus uniform criteria should be defined to consolidate BPE as a biomarker. 31 Furthermore, in the era of new cellular signaling pathways and molecular therapies, CE-MRI can be used for quantitative assessment of the vascular microenvironment and the tissue permeability. 9 Breast lesions kinetic patterns differ between malignant and benign lesions; thus, enhancement time/intensity curve characteristics can be used in combination with morphologic features to improve differential diagnosis. Semi-quantitative parameters can be extracted from the enhancement curves, including the onset time, maximum signal intensity, gradient or rate of contrast uptake and washout, and initial area under the time signal curve. 32,33 In recent years, the associations between contrast-enhancement kinetics and molecular subtypes were widely investigated. 32,34 According to Blaschke and Abe, 32 HER2 positive tumors demonstrated a faster and earlier enhancement than other subtypes, while luminal A and basal cancers showed a reduced washout during the delayed phase. This can be attributed to the frequent association of luminal A cancers with ductal carcinoma in situ, which rarely demonstrates washout kinetics; while basal subtype cancers are often characterized by tumoral necrosis and central scarring, which typically shows a persistent enhancement. 32 Quantitative analysis involves pharmacokinetic modeling and requires more complex methods for estimating changes in tissue contrast agent concentration following intravenous injection. The transfer constant, K trans , describes the transendothelial transport of contrast medium by diffusion from the vascular space to the tumor interstitium and provides a measure of vascular permeability. Gradually, gadolinium diffuses back into the vascular system, with K ep representing the transfer constant from the interstitium to blood plasma and Ve the extravascular-extracellular volume fraction. K trans and K ep are generally high in tumors. A significant reduction up to a third has been detected in both parameters in patients with locally advanced BC early responding to NAT, 35 while an increase in Ve has been shown in non-responders. 36 According to the authors O'Flynn and Nandita M. de Souza, 33 K trans can be used as a predictive biomarker to evaluate response to antiangiogenic drugs or vascular disruptive agents such as bevacizumab, a humanized monoclonal antibody directed against the vascular endothelial growth factor (VEGF), with a change in K trans value >40% commonly considered as the threshold for definitive disease response. 37 In summary, the available literature shows an ample consensus on the diagnostic value of CE-MRI measurements for non-invasive characterization and prognostication of BCs as well as for therapy monitoring during NAT. DWI A review of the literature emphasizes DWI as a potential source of biomarkers to increase breast MRI specificity, significantly improving diagnostic accuracy and reducing unnecessary biopsies. 38 DWI explores different functional tissue features including water molecules motion in the extracellular space, density of neoplastic cells, tissue microstructure, cell wall 43 Authors demonstrated that ADC value ≤ 1.01 10 −3 mm 2 /s allows the identification of invasive tumors with 78% sensitivity and 90% specificity (Figure 2). In the same research, no significant differences in ADC values were found between high-and low-grade tumors, in contrast to a previous study in which a correlation between high histopathological grade (G3) and low median ADC values was found. 39 Subsequently, a significant association was found between high ADC values and luminal A subtype. 40 Guvenc et al described a correlation between low ADC values and more aggressive subtypes of BC, secondary to high cell density. 42 In particular, a statistically significant relationship was found between low ADC values and low hormone receptors positivity along with the presence of abnormal lymphnodes. In 2007, Hamstra et al 44 first introduced DWI as a biomarker to assess the response to NAT in different types of cancer, including BC. Significant preclinical and clinical studies were performed to support the hypothesis that DWI was a promising biomarker for early evaluation of response to NAT. ADC values variations may give early information regarding response to therapy, due to ADC peculiarity to reflect tumor cellularity and necrosis status. 45,46 Park et al 47 found an association between BCs pre-treatment low ADC value and better response to chemotherapy. The accuracy of ADC in predicting the response to NAT was evaluated by Richards et al 48 who concluded that pre-treatment tumor ADC values varied according to breast tumor phenotypes and were predictive of pathologic response in TNBCs ( Figure 3). [49][50][51] However, the wide variability of results in the literature and the lack of standardization are two major limitations of DWI and DWI-derived biomarkers. To overcome these drawbacks, the European Society of Breast Imaging (EUSOBI) has established a multicenter, international working group composed of clinical experts, MRI physicists and MRI equipment suppliers with experience in breast DWI. 52 DWI working group objectives include the promotion of DWI in MRI protocols, the diffusion of technical guidance for DWI protocols and the creation and improvement of quality control methods, to finally find agreement on the optimal image processing, visualization and interpretation. In a recent review, Iima et al addressed advanced DWI models, such as intravoxel incoherent motion (IVIM), diffusion tensor imaging (DTI) and diffusion weighted kurtosis (DKI). 38 IVIM is a biexponential model that simultaneously evaluates BJR\|Open Review article: MRI biomarker for breast cancer personalized management tissue diffusivity and tissue microcapillary perfusion. DTI gives quantitative data on the water molecules directional diffusivity in biological tissues. The information obtained about diffusion anisotropy could be a potential biomarker of malignancy. It has been hypothesized that proliferating neoplastic cells, which generally destroy the normal structure of the mammary gland, could reduce anisotropy. DKI quantifies the incoherent movement of water molecules and tissue microperfusion typical of non-Gaussian phenomena, useful in the detection and characterization of breast lesions. A model proposed to quantify the Gaussian and non-Gaussian diffusion is able to estimate the D value, which represents the Gaussian diffusion, and the K value, a kurtosis parameter that represents the deviation from the Gaussian diffusion. These technological advances are supported by several studies and revealed to be useful in establish benign or malignant nature of breast lesions, in evaluating Ki-67 and tumor grading and in predicting treatment response. DWI is a promising qualitative and quantitative biomarker, a valid tool in lesion characterization and therapy monitoring. However, standardization of the acquisition and interpretation modalities of the extracted DWI data will enhance its clinical value. MRS MRS is a non-invasive functional technique that provides information on biochemical changes in specific anatomic structures by identifying and monitoring the chemical composition of the tissue under examination. In the last decades, proton MR spectroscopy ( 1 H-MRS) based on the detection of the total choline peak (tCho) has been implemented in breast Mp-MRI protocols, since several studies reported higher levels of tCho in BCs compared to benign lesions and normal breast tissue. 53,54 Therefore, on the basis of different choline peaks, MRS is a potential biomarker to improve diagnostic accuracy and predict tumor aggressiveness. 55 The diagnostic accuracy of a high-spatial-resolution 3D 1 H-MRS protocol at 3 T was evaluated for the characterization of breast lesions, based on tCho signal-to-noise ratio threshold levels and proved its potential in becoming a valuable adjunct to CE-MRI in distinguishing between benign and malignant breast lesions. 56 Other authors demonstrated that tumor tCho measurements were significantly higher in invasive ductal carcinomas vs in situ cancers and that tCho correlated with numerous prognostic factors, including histologic and nuclear grade, and estrogen receptor status. 57 Thus, the addiction of MRS in multiparametric protocol leads to significantly higher diagnostic accuracy than CE-MRI, also significantly reducing false-negative and falsepositive cases. 58 However, tCho sensitivity significantly decreases for smaller cancers, due to insufficient detection of tCho signal. 59 Available data suggest tCho as a potential biomarker for treatment response assessment and early prediction of the final NAT effect. In treated lesions, an early decrease in tCho levels, after the initial course of therapy, is consistent with tumor response and is even more sensitive than other morphological and functional criteria. Instead, it was demonstrated an increase in tCho concentration in patients with local recurrence. [60][61][62] Beyond tCho further metabolites can be detected and monitored, above all, the most promising seems to be the assessment of lipid metabolism. Thakur et al demonstrated that intratumoral lipid concentration allows to distinguish benign from malignant tumors and to differentiate among BC molecular subtype. 63 Ramadan and colleagues 64 described that healthy breast tissue in patients with BRCA-1 and BRCA-2 mutation was likely to differ from non-mutation carriers in unsaturated fatty acids triglycerides and cholesterol levels. Further studies are needed to support these results, that could open new scenarios in high-risk females screening programs. Recently, phosphorus spectroscopy ( 31 P MRS) has been introduced as a new functional MRI parameter for BC diagnosis and therapy monitoring. In particular, it has been demonstrated that a decrease in the phosphoethanolamine/phosphocholine (PE/PC) ratio is a sensitive cellular-level indicator of malignancy. 65 Moreover, changes in PE/PC ratios are observed during NAT. These promising results of 31 P MRS have been obtained with 7 T-scanners. Thus, due to the lack of data collected with ultra-high field scanners certainly related to their limited diffusion, the use of 31 P MRS as a breast MRI biomarker is still limited in clinical practice. ARTIFICIAL INTELLIGENCE AI is a computer science branch able to analyze a multitude of complex data. In recent years, AI potential has been exploited in diagnosis, treatment and outcome prediction of many clinical conditions, including BC. Breast MRI, due its intrinsic multiparametric concept, is inherently suitable for AI applications. Each breast MRI generates multiple volumes of images that can be integrated and arranged according to the different diagnostic, therapeutic or prognostic purposes. 66 CE-, T 1 -and T 2 weighted, DWI and MRS images provide large datasets fitting AI applications and potential MRI biomarkers. In Gilles et al landmark paper was clearly stated that "images are more than pictures, they are data", focusing on the hidden power of imaging, including information not always perceptible by human eye. 67,68 The term "radiomics" was first used in 2010, to describe the process of building predictive models via quantitative data extracted from radiological examinations. Radiomics consists of different stages, which includes image acquisition, reconstruction, segmentation and rendering, features extraction and qualification, database and data sharing for any ad hoc computer analysis. 69 The goal of radiomics is to provide valuable diagnostic, prognostic or predictive information derived from biological and medical data. 70 In a review published in 2018, radiomics models based on different imaging methods including MRI were investigated. Studies that analyzed BC using a radiomic approach and that provided data on BC diagnosis (detection or characterization), BC prognosis (response to therapy, morbidity, mortality) or provided data on technical challenges (software application: open source, repeatability results) were included. Authors concluded that the application of radiomics in BC patients was an emerging translational research topic, with the capability of improving the knowledge of the breast lesions specifics. 71 Currently, radiomics encounters many obstacles: the need for large clinical data and standardized protocols, the dispersion of data in different centers, the excessive costs for technological development. In recent years, several countries have already adopted many approaches to control variability in clinical trial protocols, data acquisition and analysis. For instance, across Europe, consistent protocol guidance was achieved with the help of European Association of Nuclear Medicine. The Quantitative Imaging Biomarker Alliance initiative also aims to reach the same task in a much broader level. The known clinically significant genetic variables in BC and the good time and spatial resolution of breast MRI provide an excellent basis for radiogenomics research. 72 At present, breast imaging radiogenomics has primarily centered on CE-MRI sequences, focusing on differentiation of molecular subtypes and assessment of recurrences. 73 Several authors identified an association between different radiomic tumor phenotypes and various genomic features involved in multiple layers of molecular regulation and gene expression profiles of BCs. 74,75 Other authors 76 investigated possible correlation between imaging features and prognostic genomic tests such as Oncotype Dx, concluding that breast MRI has the potential to serve as a source of imaging biomarkers in the prediction of cancer recurrence. Further studies demonstrated a correlation between the expression of some genetic markers and the MRI variables during NAT although these results are still preliminary. A recent literature review 77 found that radiogenomics, combining genomic information with emerging deep learning (DL) modalities, could predict the effectiveness of NAT and provide information on disease progression. Among AI applications specific methodologies are machine learning (ML) and DL. 66 ML is a technology that allows the automatic training of machines with the aim of obtaining predictive data set based on the data and algorithms provided, without explicit programming. DL, a subset of ML, is characterized by a high accuracy, less need of human intervention but requires a huge amount of training data and expensive hardware and software. 78 Due to MRI intrinsic multiparametric nature, ML application in breast MRI is fast-paced developing and many studies are demonstrating ML usefulness in lesion detection and classification, prediction of NAT response and recurrence risk, and therefore to guide therapeutic decisions. 79,80 DL demonstrated high diagnostic accuracy to differentiate benign from malignant lesions, 79 improving the diagnostic performance of breast MRI by decreasing the false positives and improving the positive-predictive value. 81 Furthermore, DL has also been exploited extensively for evaluating the response to NAT. 77 The integration of AI into breast imaging may enable the creation of new imaging biomarkers that incorporate patient clinical and tumor structural characteristics. Moreover, biomarkers could be incorporated into patient risk stratification via personalized imaging. 82 Nevertheless, AI applications are not ready to be incorporated into clinical practice, nor to replace expert human observers with the ability to critically evaluate MRI images and patient history. CONCLUSIONS Breast MRI may act as a diagnostic and prognostic tool to improve BC management through the extraction of a plenty of functional cancer parameters serving as imaging biomarkers. The intrinsic multiparametric nature of MRI provides specific information to visualize and quantify the functional processes of cancer development and progression, in order to improve detection and characterization of breast lesions, monitoring and prediction of response to therapy, and differentiation of biological BC subtypes. Moreover MRI images, due to their complex information content, are a fertile ground for AI applications. These may improve the integration of imaging biomarkers in clinical decision-making through the building of accessible predictive integrated models aiming at individualized medicine. CONFLICTS OF INTEREST The Authors declare that there is no conflict of interest. FUNDING This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.	2022-05-14T15:17:30.058Z	2022-05-12T00:00:00.000	{ "year": 2022, "sha1": "656866d0f2635cc5bb557c80fe0695ed91d8c858", "oa_license": "CCBY", "oa_url": "https://doi.org/10.1259/bjro.20220002", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "0b4f41dd08faea57cfae097ab54827db5d9ff4bb", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
214213331	pes2o/s2orc	v3-fos-license	Topology controlled phase coherence and quantum fluctuations in superconducting nanowires Superconducting properties of metallic nano-wires may strongly depend on specific experimental conditions. Here we consider a setup where superconducting phase fluctuations are restricted at one point inside the wire and equilibrium supercurrent flows along the wire segment of an arbitrary length $L$. Low temperature physics of this structure is essentially determined, on one hand, by smooth phase fluctuations and, on the other hand, by quantum phase slips. The zero temperature phase diagram is controlled by the wire cross section and consists of a truly superconducting phase and two different phases where superconductivity can be observed only at shorter length scales. One of the latter phases exhibits more robust short-scale superconductivity whereas another one demonstrates a power-law decay of the supercurrent with increasing $L$ already at relatively short scales. Introduction Superconducting properties of ultrathin nanowires are markedly different from those of their bulk counterparts. The key reason for that lies in the presence of fluctuation effects which become particularly pronounced in quasi-one-dimensional superconducting systems and persist down to T → 0 [1,2,3]. Fluctuations in superconducting nanowires become progressively stronger as the wire cross section s decreases. An important parameter controlling the strength of fluctuation effects is g ξ = R q /R ξ ∝ s, where R q = 2π/e 2 stands for the quantum resistance unit and R ξ defines the resistance of a wire segment of length equal to the superconducting coherence length ξ. Gaussian fluctuations of the superconducting order parameter give rise to a negative correction δ∆ ∼ ∆/g ξ to the value of the superconducting gap ∆ [4]. The effects caused by strong (non-gaussian) fluctuations of the order parameter -the so-called phase slips -turn out to be even more significant. A phase slip can be viewed as localized both in space and in time full suppression of the absolute value of the order parameter accompanied by the phase jump by ±2π. At temperatures outside an immediate vicinity of the critical one T C such fluctuations have a quantum origin and, hence, are usually referred to as quantum phase slips (QPS). They can also be thought of as two-dimensional tunneling processes with the amplitude [5] γ QP S ∼ (g ξ ∆/ξ) exp(−ag ξ ), a ∼ 1. (1) Compelling experimental evidence for the presence of QPS effects in superconducting nanowires was provided in a number of works [6,7,8]. Another dimensionless parameter g = R q /Z w ∝ √ s specifically accounts for the effects associated with long range fluctuations of the superconducting phase ϕ. Here Z w = L kin /C w is an effective impedance of the superconducting wire which can be viewed as a transmission line with (kinetic) inductance L kin and (geometric) capacitance C w . Note that smooth phase fluctuations are associated with sound-like plasma modes propagating along the wire, the so-called Mooij-Schön modes [9]. Among the phenomena caused by such fluctuations is the effect of smearing of the electron density of states (DOS) [10] which was recently verified experimentally [11]. Being controlled by g, this smearing can occur even at g ξ ≫ 1 implying both the presence of subgap electron states at any finite temperature and an effective suppression of the gap edge singularity down to its complete elimination at g = 2 even at T = 0. The magnitude of logarithmic inter-QPS interactions is also controlled by the parameter g. For g > 16 this interaction remains strong and, hence, "positive" and "negative" QPS remain bound in pairs. In this regime the phase coherence inside the wire is essentially preserved and the wire exhibits vanishing linear resistance. Accordingly, this state can be considered superconducting. For g < 16, on the contrary, the inter-QPS interaction gets weaker and unbound QPS appear as a result of Berezinskii-Kosterlitz-Thouless-like (BKT) quantum phase transition (QPT) at g = 16. In this case the wire acquires a non-zero resistance [12], thus signaling the absence of a superconducting response. Note, however, that such a behavior, can be observed in a specific type of experiments while in different setups the wire can exhibit superconducting properties even at g < 16 [13]. A somewhat similar situation was earlier discussed in details in the case of one-dimensional arrays of resistively shunted Josephson junctions [14]. Yet another type of experiment was recently considered in [15]. The corresponding setup enables passing equilibrium supercurrent across an arbitrary segment of the wire without restricting fluctuations of its superconducting phase. It was demonstrated that the behavior of a superconducting nanowire with g < 16 is determined by the bath of collective plasma modes and turns out to be richer than that identified earlier for different experimental setups [12,13]. In particular, for the configuration analyzed in [15] total suppression of the supercurrent occurs only for g < 2, while at 2 < g < 16 the wire exhibits a mixed behavior with two different correlation lengths. This phase turns out to be superconducting at shorter length scales and non-superconducting at longer ones. In the present work we will further investigate equilibrium properties of a "disordered" phase with g < 16 and consider the setup that allows to effectively restrict the space available for phase fluctuations by "pinning" Fig. 1 The system under consideration the superconducting phase ϕ at one point inside the wire. We will demonstrate that such topology controlled phase pinning severely enhances the ability of the wire to conduct supercurrent. This effect can be interpreted in terms of the absence of a massless mode responsible for the destruction of superconductivity at g < 2 in the setup considered in [15]. Instead, the nanowire embedded in our present setup exhibits a transition between "more" and "less" superconducting phases characterized by different types of long-range behavior. The structure of the paper is as follows. In Sec. 2 we describe the system under consideration and present our theoretical approach. In Sec. 3 we analyze the influence of smooth phase fluctuations on the supercurrent. Sec.4 is devoted to the effects caused by QPS. Finally, in Sec. 5 we discuss our findings and compare them to previous results. Model and formalism In what follows we will restrict our attention to the setup displayed in Fig. 1. It features a long and sufficiently thin superconducting nanowire with a bulk superconducting reservoir attached to one of its ends. This reservoir has a form of an open ring whose opposite end is attached to the wire by a small-area tunnel junction at a distance L along the wire. The open ring is pierced by an external magnetic flux Φ which controls the phase difference φ = 2πΦ/Φ 0 between its ends. Accordingly, the phase at the left end of the wire is pinned by the reservoir and is set equal to zero, i.e. ϕ(x = 0) = 0. We assume our system to be in thermodynamic equilibrium at T → 0. Our setup allows one to investigate the ability of the wire to conduct non-vanishing supercurrent in a specific phase-biased measurement complimentary to those analyzed elsewhere [12,13,15]. Our observable of interest is the electric current I(φ) flowing through the wire segment of length L between the left wire end and the junction. As the supercurrent I is a 2π-periodic function of the phase φ we will restrict the phase φ to the interval φ ∈ (−π, π). Low-energy physics of the system is conveniently described by the imaginary-time effective action Here and below the imaginary time variable τ is omitted for brevity. The first term in this formula represents the wire effective action [5,12,16] while the S J -term accounts for the Josephson coupling energy and reads Integrating out the phase variable ϕ(x) at all points along the wire except for its value at x = L we arrive at the reduced effective action which depends only on the phase ϕ(L, τ ) ≡ ϕ: and v = e 2 g/(2πC w ) being respectively the bare Matsubara propagator of ϕ and the velocity of the Mooij-Schön mode. We observe that fluctuations of the phase variable are massive with m 0 = gv/8πL. The absence of a massless mode in our setup (in contrast to that considered in [15]) is a direct consequence of phase pinning at x = 0 which prohibits uniform shifts of the phase inside the wire. Following our previous work [15] we are going to employ a variational technique in the form of a selfconsistent harmonic approximation (SCHA). More details related to this method can be found, e.g., in Ref. [17]. Let us define the trial action The variational parameter m accounts for the interactioninduced effective mass for the ϕ-mode and ψ represents the average value of the phase difference. Evaluating the free energy of the system as a function of these two parameters and minimizing it with respect to both m and ψ, we arrive at the following SCHA equations Here and below we define where ω k = πT (2k + 1) is the Matsubara frequency. From Eqs. (8), (9) we observe that within our variational approach the effect of phase fluctuations is fully accounted for by effective renormalization of E J by the factor e −G(0)/2 . The supercurrent I is then found from the equation Supercurrent in the presence of fluctuations Let us now make use of Eqs. (8)- (11) and explicitly evaluate the zero temperature supercurrent inside the wire segment of length L, see Fig. 1. At L v/∆ phase fluctuations are strongly suppressed and the system remains in the mean-field regime. In the opposite limit of large L the solution of Eq.(8) exhibits two qualitatively distinct regimes separated by g * = 4. At g < 4 we find \|m\| ≪ v/L. Therefore, the emergent mass is negligible, and the effect of fluctuations is purely gaussian. The equation of motion (9) is then rewritten as In the interesting for us limit of small E J we may readily set 8πEJ g∆ < 1. In this case the sine term is renormalized to zero faster than the kinetic inductance contribution ∝ L −1 and, hence, we obtain This expression demonstrates that for g < 4 phase fluctuations (i) modify the current-phase relation making it sine-like instead of the sawtooth-like (the latter is realized in the long L limit mean-field regime) and (ii) yield a decrease of the supercurrent as compared to the standard Josephson formula I(φ) = 2eE J sin φ that applies in the limit L → 0. In addition, we observe that in the presence of fluctuations the supercurrent (13) decays faster with increasing L than the standard mean field dependence I ∝ 1/L. Let us now turn to the case g > 4. Resolving Eq. (8) in the limit L → ∞ we obtain This solution remains valid only as long as L exceeds the new length scale of our problem L * which reads This length scale separates the regime L > L * where fluctuations lead to a non-gaussian renormalization of the interaction potential from the gaussian regime L ≪ L * where \|m\| ≪ gv/8πL. As long as v/∆ ≪ L ≪ L * the current is again given by Eq. (13). For g > 4 the renormalized Josephson coupling energy decreases slower than 1/L and at L ∼ L * it becomes of the same order as the kinetic inductance contribution. At even larger distances the mass renormalization saturates to the value defined in Eq. (14). The kinetic inductance contribution, on the contrary, decreases as 1/L. Therefore, at L ≫ L * the phase is pinned to the lowest minimum of the renormalized Josephson junction potential, i.e. we have ψ = φ. In this case the current-phase relation reduces to the standard mean field form The effect of QPS The above analysis accounts only for the effect of smooth phase fluctuations and does not include quantum phase slips. Their influence is essentially identical to that already analyzed in Refs. [13,15] for different setups. Hence, at this point it suffices to only briefly summarize our key observations. As we already discussed, for g > 16 "positive" and "negative" quantum phase slips are bound in pairs thus causing no long-range phase decoherence. Hence, for such values of g all results of the previous section remain applicable also in the presence of QPS. In contrast, at g < 16 quantum phase slips are no longer bound in pairs. In this case yet another length scale appears in our problem. It plays the role of a decoherence length due to QPS and reads [13,15] As L grows larger, quantum phase slips yield stronger disruption of long range phase coherence along the wire and, hence, faster decay of the supercurrent. In particular, at L ≫ L c we have [13] Combining this result with those derived above in the previous section we arrive at the following physical picture. For g < 4 there exists only one correlation length (17) in our problem. At L ≪ L c QPS effects are irrelevant and the supercurrent suppression is merely due to smooth phase fluctuations. In this limit Eq. (13) applies and the supercurrent decays as a power-law with increasing L. As soon as L exceeds L c quantum phase slips come into play and the supercurrent decay becomes exponential with L, as it is seen in Eq. (18). Thus, in practical terms the wire loses its ability to carry supercurrent in the limit of large L ≫ L c . The situation becomes somewhat more complicated for 4 < g < 16, since in this case there exist two different correlation lengths in our problem, L * and L c . The first one diverges as g → 4 while the second one tends to infinity at g → 16. Depending on the relation between these two lengths, a number of different regimes can occur. Let us first consider the limit L * ≪ L c which can always be realized for sufficiently large values of g ξ . As before, at shorter length scales L < L * only smooth phase fluctuations affect the supercurrent causing its powerlaw suppression with increasing L and the sinusoidal current-phase relation, see Eq. (13). At L * < L < L c both smooth phase fluctuations and quantum phase slips are practically irrelevant and the supercurrent is defined by the standard mean field result I ∝ 1/L (16) describing the sawtooth-shaped current-phase relation. Finally, at L ≫ L c the current is exponentially suppressed and the current-phase relation again reduces to the sine form, see Eq. (18). For certain values of the system parameters -in particular for g close to 4 -one can also realize the regime L * > L c . In this case there exists no room for the mean field result (16), whereas both Eqs. (13) and (18) remain applicable in the corresponding limits. Discussion Superconducting properties of metallic nanowires depend not only on their parameters, but also on the topology of the experimental setup and on the way the experiment is being performed. The ability of the wire to carry supercurrent also varies at different length scales being affected by different kinds of fluctuations. In this work we investigated how fluctuations affect supercurrent in a long superconducting nanowire as a part of the setup displayed in Fig. 1. This setup allows the wire to pass an equilibrium supercurrent across a segment of arbitrary length L driven by an external magnetic flux. Configurations of phase inside the wire are restricted only by a bulk reservoir which pins the superconducting phase at one of the wire ends. Fluctuation effects manifest themselves, on one hand, via sound-like collective plasma modes forming a quantum dissipative environment for electrons inside the wire and, on the other hand, via quantum phase slips. The effect of a dissipative environment formed by plasma modes boils down to an effective renormalization (reduction) of the Josephson current. As the bath of collective modes is almost Ohmic (having a lower cut-off frequency which scales as 1/L) the low temperature system behavior resembles that involving the so-called Schmid dissipative QPT [18]. For g smaller than the critical value (g < 4) the effect of smooth phase fluctuations is purely gaussian. It leads to a power-law decay of the current (see Eq. (13)). The current-phase relation is sine-like in this regime. For g > 4 a new length scale L * defined in Eq.(15) appears beyond which the bath obtains a finite interaction-induced mass as in Eq. (14) and the phase becomes pinned to the value defined by an external magnetic flux. At this scale the current becomes insensitive to smooth phase fluctuations and is given by a simple mean field formula (16) with a sawtooth shape of current-phase dependence. The presence of quantum phase slips naturally leads to a BKT-type quantum phase transition at g = 16 [12]. For g > 16 QPS do not have any significant impact on the supercurrent and the wire retains superconductivity at any length scale (with supercurrent somewhat reduced by fluctuations at L < L * ). At g < 16 phase slips are unbound and cause an exponential decay of the current at L ≫ L c according to Eq. (18) where the phase relaxation length L c is defined in Eq. (17). Smooth phase fluctuations are irrelevant in this regime and the wire loses superconductivity at soon as L strongly exceeds L c . On the contrary, at L < L c QPS do not play a significant role and the physics is completely determined by smooth phase fluctuations. It is interesting to compare our present results with those derived recently in Ref. [15]. While the setup [15] allows for unrestricted fluctuations of the superconducting phase, here we consider a different topology which effectively pins the phase at one of the wire ends. In the former case a gapless Ohmic mode (associated with uni-form phase shifts along the wire) appears at any L, in contrast to the situation considered here. Fluctuations associated with this gapless mode cause a Schmid-like QPT at g = 2. As a result, in the setup considered in Ref. [15] the wire completely loses superconductivity at g < 2, whereas the phase with 2 < g < 16 is mixed, i.e. it is non-superconducting in the long length limit and superconducting at shorter scales, even though the gapless mode causes additional suppression of current in the limit L → 0. Comparing this situation with the one considered here, we observe that superconductivity is severely enhanced by the phase pinning as a result of the soft mode suppression. This effect turns the QPT at g = 2 [15] into a transition between "less" and "more" superconducting phases at g = 4 considered here. We also note that a similar phase transition was also discussed in Ref. [19] in the context of superconducting nanorings interrupted by a Josephson junction. It would be interesting to verify our predictions in experiments with superconducting nanowires.	2020-01-30T09:05:01.086Z	2020-01-24T00:00:00.000	{ "year": 2020, "sha1": "b352ceb84df91b52045aef14b3af94dcf8450937", "oa_license": null, "oa_url": "http://arxiv.org/pdf/2009.01165", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "6b507c9cae6fdbc7f5f35ea862fbe80313a69503", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
253717876	pes2o/s2orc	v3-fos-license	Transferability of the working envelope approach for parameter selection and optimization in thin wall WAAM This work aims to propose and assess a methodology for parameterization for WAAM of thin walls based on a previously existing working envelope built for a basic material (parameter transferability). This work also aimed at investigating whether the working envelope approach can be used to optimize the parameterization for a target wall width in terms of arc energy (which governs microstructure and microhardness), surface finish and active deposition time. To reach the main objective, first, a reference working envelope was developed through a series of deposited walls with a plain C-Mn steel wire. Wire feed speed (WFS) and travel speed (TS) were treated as independent variables, while the geometric wall features were considered dependent variables. After validation, three combinations of WFS and TS capable of achieving the same effective wall width were deposited with a 2.25Cr-1Mo steel wire. To evaluate the parameter transferability between the two materials, the geometric features of these walls were measured and compared with the predicted values. The results showed minor deviations between the predicted and measured values. As a result, WAAM parameter selection for another material showed to be feasible after only fewer experiments (shorter time and lower resource consumption) from a working envelope previously developed. The usage of the approach to optimize parameterization was also demonstrated. For this case, lower values of WFS and TS were capable of achieving a better surface finish. However, higher WFS and TS are advantageous in terms of production time. As long as the same wall width is maintained, variations in WFS and TS do not significantly affect microstructure and microhardness. Introduction Wire arc additive manufacturing (WAAM) has gained momentum as a group of processes capable of achieving high deposition rates when compared with other AM processes applied to metals. WAAM processes also stand out for their great versatility in terms of metallic materials, ranging from high-strength low alloy (HSLA) steels [1][2][3] to Ni-rich alloys [4]. From the development side, according to Wu et al. [5], depending on the version, Gas Metal Arc (GMA) can reach deposition rates between 2 and 3 times higher than Gas Tungsten Arc (GTA) and Plasma Arc (PA). Williams et al. [6] pointed out that GMA allows for more accessible trajectory programming due to the coaxiality between wire and torch. These characteristics make GMA prominent for the fabrication of metallic components in a relatively short time. Despite the advantages, the application of GMA as a power drive in WAAM requires time and resources to find adequate operational parameters and provide proper geometric features, surface finish and material properties. From the end-user side, a body of experts for process parameterization is not desirable (out of the business core field). End-users need to satisfy their clients with short delivery times and attending custom orders. They do not want to carry out many experiments to define parameters for this specific technology. Thus, harnessing strategies for cost and time-effective parameterization is crucial for maturing this technology in terms of acceptance from industrial end-users. In this scenario, the operational map application can be pointed as a potential solution, allowing the prediction of parameter ranges capable of meeting the requirements for the mentioned aspects. This approach was successfully used in the past for welding and overlaying operations. But still not fully explored for WAAM, to the best of the authors' knowledge. In the working envelope approach, the process parameters are assumed as independent variables and are usually systematically varied, following an experimental design. According to the analysis objective, the variables of interest (dependent variables), such as geometry or discontinuities, are assessed quantitatively or qualitatively. In some cases, the construction of such maps allows for the development of equations and contour plots for predictions. Following this approach, Marinelli et al. [7] built a working envelope for autogenous welding with GTAW, evaluating different levels of travel speed and shielding gas compositions (varying Argon and Helium contents). The authors found that higher He contents in the shielding gas and lower travel speeds resulted in weld beads free of cracks. In another work, Ahsan et al. [8] investigated the viability of applying Cold Metal Transfer (CMT) GMAW to weld overlap joints without any gaps. According to the authors, two working envelopes were identified in terms of porosity and mechanical properties: one for low heat input (200 to 250 J/mm) and another for high heat inputs (350 to 550 J/mm). Dye et al. [9] conducted a numerical analysis to predict parameter combinations (effective power and travel speed) capable of avoiding lack of penetration, porosity, liquation cracks and solidification cracks for welding a nickel superalloy (IN718) with GTAW. The authors proposed a weldability map and identified a region characterized by proper weld features (working envelope) based on the results. In additive manufacturing, some developments approached operational maps, mainly when Laser is used as a power source. Thomas et al. [10], for instance, compiled literature data for different materials deposited by powder bed fusion (PBF), aiming at avoiding porosity, voids and solidification cracks. Similarly, Dass and Moridi [11] also compiled literature data for materials deposited by direct energy deposition (DED) with Laser. This latter study built a map considering heat input and powder feed rate as independent variables; three regions of non-conformity incidences (keyholing, lack of fusion and porosity) were identified. As one of the pioneers in terms of parameterization for WAAM, Martina et al. [12] developed a working envelope for the Plasma Arc to prevent the formation of different geometric irregularities in walls deposited with a Ti alloy. At the end of the study, they proposed a statistical model capable of maximizing the layer height and deposition rate for a given effective wall width, aiming to select process parameters. As can be seen, the use of working envelopes in the context of the operational maps follows the same principle, yet with different objectives. In the case of WAAM, there is the goal of achieving a determined dimension in the construction of the component. Thereby, different from other applications, including welding and overlaying, the authors of the present paper conceive that the maps for these processes should consider the wall width as a dependent variable (which is the starting point for an AM build design), whilst wire feed speed (WFS) and travel speed (TS) are considered independent variables (electric current can also be used replacing WFS). For illustration matters of the authors' proposal, Fig. 1a represents an operational map, which comprehends a global universe of parameters. The working envelope represents the region within the map where the parameter combinations provided results that meet given acceptance criteria. Figure 1b, in turn, exemplifies the parameter selection. As can be noticed, different combinations of WFS and TS (points I, II and III) can be established parameter selection plan within a working envelope to achieve the same target width (W). Hypothetically, the parameter combination labelled as "I" can be the one that results in a better surface finish. Combination "II" might be the one that guarantees a more robust product (since it is distant from the boundaries of the envelope). Combination "III" would potentially reach a shorter deposition time. Therefore, selecting one of the parameter combinations depends on each characteristic priority level. In summary, a working envelope can be used not just to find a process parameter window that avoids non-conformities (as usually this concept is applied for). However, it is also applicable for optimization targeting specific characteristics. In this context, the same research group of this proposal built a working envelope for WAAM using an Al-5Mg alloy as feedstock [13]. The acceptance criteria to delimit the working envelope took into account surface aspect, surface waviness (< 0.5 mm) and porosity formation (< 3.0 %). Complementarily, Da Silva et al. [13] introduced a model of the working envelope, visualized by contour plots for total width, effective width and layer height, as a function of WFS and TS. Also exploring the potentiality of operational maps for WAAM by the same group, Dahat et al. [14] earlier described a step-by-step methodology for the construction of working envelopes. As a case study, they used a HSLA steel deposited with CMT. By comparing the predicted dependent variables with those experimentally found, the authors proved a high reputability capable of validating the proposed methodology. Figure 2 presents a compilation of the working envelopes built by Da Silva et al. [13] and Dahat et al. [14] for an Al-5Mg alloy and an HSLA steel, respectively. As can be seen, despite the use of very distinct materials, both working envelopes present essentially the same format. However, the working envelopes assumed different positions within the operational map since the boundaries changed according to the material and other process variables. Examples of essential variables in this context can be shielding gas, wire diameter, contact tube to work distance (CTWD), interlayer temperature, thermal management method (forced cooling) and even arbitrary acceptance criteria adopted to construct the maps. Nevertheless, this finding indicates that it could be possible to use a pre-existing envelope to find parameters for another material. As evidenced by Da Silva et al. [13] and Dahat et al. [14], the working envelope approach shows itself as a robust and practical tool for parameter selection in thin WAAM. The concept of thin walls considers single beads per layer, with no torch oscillation in the transverse direction. However, it still demands experimental work (time and resource consuming) to be raised. Aiming at making this approach more cost-effective and functional, the main objective of this work is to verify the possibility of parameterizing thin walls deposited by WAAM for a feedstock, based on a pre-existing working envelope made with a different and ordinary (cheaper) material. As a complementary objective, it aims at investigating whether or not the working envelope can be used to optimize parameter selection of a target wall width, in terms of operational characteristics (objective functions), such as surface finish, active deposition time, arc energy (which governs microstructure and microhardness). Here, the surface finish is understood as that resulted from the deposition, with no after work (machining). As a whole, this proposal tries to contribute to maturing a not widely used methodology for WAAM parameter selection for thin wall WAAM. Methodology and experimental procedures A working envelope employing a C-Mn steel wire (1.2-mmdiameter wire, AWS ER70-S6 class - Table 1), hereafter named as "reference working envelope", was developed as a base (pre-existing working envelope) to achieve the main objective. As is well-known, this class of material is trendy in welding constructions, and users dominate the technical application. Moreover, this feedstock is reasonably cheap. Then, it seems to be an ideal material for the "reference working envelope" for structural steels, which demand more experiments to be built. This proposed working envelope is applicable to build multilayered wall-like parts deposited with a single pass per layer. The depositions were carried out using a CNC gantry machine coupled with a Fronius power source. A schematic illustration of the employed experimental rig is shown in Fig. 3. The GMA variant Cold Metal Transfer (CMT) was used as a depositing process (synergic line code CMT 0963). The wire feed speed (WFS) and travel speed (TS) were assumed as independent set variables. They were systematically varied, whilst the geometrical features (external and effective width, layer height and surface waviness) were defined as dependent response variables. Subsequently, validation of the reference envelope was implemented using three combinations of set WFS and TS, targeting the same effective width within the working envelope. The geometry of the walls built in the validation step was analysed and compared with the predictions from the working envelope approach. Besides this, cross-sections of the walls were taken and prepared to verify possible changes in microstructure and microhardness. Once the reference working envelope had been built, characterized and validated, the parameterization transferability from the reference envelope to another material was evaluated. To do so, only three new walls were deposited with a 2.25Cr-1Mo steel (1.2-mm-diameter wire, AWS ER90S-B3 class - Table 1), using the same set-up and parameters employed in the combinations for the reference envelope validation. The potentiality of transferring the parameters from the reference working envelope to this other material was assessed by comparing the measured results with the predictions. Similarly, the microstructure and microhardness from the 2.25Cr-1Mo deposits were also analyzed. Although both wires used in this work are structural carbon steels, it must be highlighted that these materials have different costs and applications. The low alloy carbon steel, referred here as 2.25Cr-1Mo (AWS ER90S-B3), can be up to four times more expensive. This steel corresponds to a high strength low alloy (HSLA) steel that has functionality in, for instance, the oil and gas industry, where it is often employed to manufacture flanges and fittings that operate at high temperatures. Besides, since the success of the parameterization transferability is highly dependent on the physicochemical compatibility between materials, working with dissimilar classes of feedstock was not an option in this methodology. The same procedure proposed by Dahat et al. [14] was followed to build the reference envelope and the walls for the validation/transferability trials. In this procedure, substrates, made of cold-rolled steel bars, were clamped in a fixture with the narrower side facing up. This assembly format aimed at simulating a previously built wall (hereafter referred to as "pre-wall"), with the same width as the wall to be built. All walls reached a minimum deposition height of 40 mm, although the number of layers depended on the parameters for each case. For all the depositions, the shielding gas utilized was a mixture of 96% Ar + 4% CO 2 , at a 15 L/min flow rate. The contact tube-to-work distance (CTWD) was also kept constant, at 16 mm. An infrared pyrometer was used to verify the interlayer temperature, so that this temperature would be kept at around 30 °C for all depositions, regardless of the purpose of the built walls (reference envelope construction, trials for envelope validation or trials for parameter transferability). The interlayer temperature verifications were always carried out at the longitudinal centre point of the top surface of each previously deposited layer. The electric signals (current and voltage) and wire feed speed were monitored in each deposited layer, through an A/D board, at 5 kHz rate and during 8 s. Mean and root mean square (RMS) values of current and voltage were calculated for each wall produced, discarding (6) motor drivers arc start and arc ending data acquisition regions. Average wire feed speed, mean current (I m ), RMS current (I rms ), mean voltage (U m ), RMS voltage (U rms ) and arc energy per unit of length were calculated for 10 layers in each wall (10 averages of each mean quantity calculated through the acquisition times in each layer). Arc energy per unit of length was calculated by computing the average instantaneous power (average of the point-by-point product of current and voltage) divided by TS. As already mentioned in the literature [15,16], the control strategy used by the CMT and its variants lead to differences between set and actual wire feed speeds. Due to this equipment performance, the average wire feed speed values were measured with a properly calibrated encoder (0.1 m/min resolution) attached to the wire feeder, and, for each deposition condition, the set WFS (WFS set ) was adjusted to reach the desired actual WFS (WFS target ). Experimental planning for building the reference working envelope Although there is a direct relationship between wire feed speed (WFS) and mean current (I m ) in GMA, the CMT welding equipment does not follow the same relation throughout the whole operational range. Thus, the working envelope inside the operational map was built as a function of WFS instead of I m (more commonly used). Moreover, since set directly in the power source interface, WFS is a more accessible parameter for the operator when compared to I m , which is a consequence of the WFS together with other variables. A theory described by Yehorov et al. [17] was considered to determine the operating ranges of WFS. The authors claim that high arc pressure should be avoided to prevent the molten pool from running down during thin wall deposition. Thus, lower current levels, still capable of guaranteeing the coalescence between layers, should be privileged in parameterizations. Therefore, preliminary tests were carried out to set WFS so that the I m values do not surpass 170 A. The three levels of target WFS (WFS target ) were 3, 4 and 5 m/min, which resulted in I m values of around 120, 145 and 170 A, respectively. To determine the travel speed (TS) range for each WFS target , two acceptance criteria were used. The criteria were based on the surface aspects of the walls deposited in preliminary tests: top surface humps and lateral sagging. Periodic humps throughout the longitudinal direction of the beads usually occur when exceeding TS value is reached (upper range limit). The lower limit of TS was defined based on the slowest possible speed that could be used without lateral sagging. To minimize irregularities and establish more conservative limits, the found lower and upper limits of TS were incremented and decremented in 5.0 cm/ min, respectively. The higher the WFS level, the higher the current (arc pressure) and the deposition rate (molten pool volume). Therefore, the lower limits of TS were not the same at each WFS set, increasing according to each level to avoid lateral sagging. The upper limits of TS, in turn, which prevents humping formation, also increased. Yuan et al. [18] claimed that the humping formation is correlated to a strong molten metal flow with high momentum, mainly affected by arc pressure, electromagnetic force and Marangoni force. As discussed by the authors, the increase in current (due to WFS) leads to a higher arc pressure and, consequently, to a greater impulse on the metal flow, facilitating the humping formation. However, a deeper melt pool is also obtained, which may dissipate the metal flow and, hence, the humps. This statement was based on their results, which showed that a short and deep molten pool is less prone to this defect than a long and shallow pool. Thus, since the TS upper limits were increased with the WFS level, it was assumed in this current work that the metal flow dissipation effect due to a deeper pool was predominant. The experimental planning matrix to build the reference working envelope for the AWS ER70S-6 wire, taken from the above considerations, is presented in Table 2. To ensure uniform distribution within each of the WFS operating ranges, the four selected levels of TS were equally divided. The experimental design for the combinations of WFS and TS will lead naturally to different wall thicknesses. To satisfy the methodological proposal, substrates with the same width as the wall to be built were used and positioned with the narrower side facing up (as afore seen in Fig. 3). However, commercial bars covering all wall widths are not available. Trying to maintain the heat flow the more uniform as possible throughout the deposited layers, the difference in width between the substrate and the wall must be the lowest possible. One solution would be to machine down each bar to the desired thickness. However, alternatively, few layers with intermediate widths were deposited over the substrates face before the actual walls, following the same strategy adopted by Dahat et al. [14], including a theoretical approximation proposed to predict the wall widths as a function of the deposition parameters. Therefore, only two different widths of cold-rolled steel bars, 150 x 50 x 7.9 mm and 150 x 50 x 6.3 mm, were employed as substrates. The commercial bar widths were selected to minimize the number of intermediate layers required, aiming at the smallest possible difference between substrate and wall widths. The following criteria were considered to deposit intermediate layers: • A difference in width up to 1 mm between substrate and wall could be capable of maintaining a constant heat flow. • When the difference between substrate and wall width predicted was smaller than 1 mm, no intermediate layer was deposited. • When this difference was between 1 and 2 mm, one intermediate layer was deposited. • When this difference was between 2 and 3 mm, two intermediate layers were deposited. Determination of the geometrical wall features All the deposited walls were digitalized through a metrology-grade 3D scanner (HandySCAN 3D TM ). The three geometrical features were measured based on the digital mesh of each wall, via dedicated software (VXElements). They are the external wall width (WW ext ), which corresponds to the broadest distance found between the wall sides, the effective wall width (WW eff ), which is the smallest distance between the wall sides, and the surface waviness (SW) calculated as the difference between WW ext and WW eff divided by two. For better sampling, the wall sides were split into two meshes and point-by-point measurements of distance between the two surfaces were taken. Figure 4 illustrates a schematic of the procedure used to quantify the geometrical features. The wall ends (arc start and arc ending regions) were discarded (Fig. 4a), since they tend to be unstable regions. Thus, only a central part of the walls, however long enough (28 x 90 mm), was considered for measurement. Based on the analysis of the WFS signals, a few regions with significantly deviated values was observed, probably due to the control made by CMT to compensate for variations in arc length caused by irregularities throughout the deposition of a layer. To avoid the influence of such non-common regions in the geometry assessment, the aforementioned central regions of the walls were divided into three equally spaced slices (28 x 30 mm each) along the wall length (Fig. 4b) and some outliers resultant from these WFS variations were neglected. In Fig. 4c, for instance, the external width value of 9.5 mm was not considered, since it corresponds to an outlier. In this way, for the given example, the largest width registered was 8.6 mm. Based on this methodology for each slice, a single value of WW ext , WW eff and SW was taken, representing the corresponding average value of each measurement. The layer heights (LH) were quantified by measuring the total wall heights with a Vernier calliper, at five different positions, and then dividing by the number of deposited layers. Additionally, chemical composition analysis was carried out on different samples with a fluorescence x-ray spectrometer (XRF-Olympus Vanta C series). Results and discussions 3.1 Building of the reference working envelope Figure 5 presents the surface aspect obtained for each of the twelve walls deposited to build the reference working envelope, using the experimental design shown in Table 2. For a same target wire feed speed (WFS target ) level, the conditions with slower actual travel speeds (TS actual ) presented poorer surface aspects (more irregularities). In this case, lower TS entails larger molten pool volumes for a same arc pressure (same current and arc length), making the weld pool more prone to lateral sagging and resulting in irregularities, corroborating the hypothesis proposed by Yehorov et al. [17]. Although lower levels of WFS (lower currents) prevent excessive arc pressure, higher levels can be used when combined with faster TS levels. Dirisu et al. [19], for example, managed to build walls with a WFS of 6.5 m/min and a TS of 40.0 cm/min, also using the CMT process and the AWS ER70S-6 wire. However, it must be taken into account that this choice could result in a narrower TS working range since the arc pressure would be higher. Table 3 presents the acquisition data from each wall deposited to build the reference working envelope. It is possible to notice that the average values of mean WFS (WFS m ) were similar for each target level, except when slower TS values were used (mainly with WFS of 5 m/min). Since walls with slower TS presented more irregular geometries, more significant variations in WFS were probably imposed by the CMT equipment to maintain constant arc length, resulting in the observed deviations. Mean current (I m ) and RMS current (I rms ) values were similar to those observed for WFS since both have a direct correlation. Moreover, mean and RMS voltage values (U m and U rms , respectively) remained at the same level for a given WFS target , independently of the TS, showing the good performance of the synergy line in keeping arc length constant. The average values of mean arc energy per unit of length (E m ), in turn, varied mainly due to the wide variations adopted for TS, and on a minor scale due to the variations in average power for a given WFS. Table 2, which attended the two acceptance criteria defined in Sect. 2.1 Table 4, in turn, presents the resultant geometrical features of the walls, namely, external wall width (WW ext ), effective wall width (WW eff ), layer height (LH) and surface waviness (SW). As seen, the standard deviations did not exceed 0.2 mm, indicating good reliability regardless of the geometric characteristic evaluated. Finally, Fig. 6a shows the working envelopes for WW ext and WW eff , whilst Fig. 6b presents the working envelope for LH, both with their respective mean values and iso-WFS target curves. Analyzing Fig. 6a for a given WFS and TS, the effective wall widths are always less than the external wall widths. As the effective width is taken in the valleys established between two layers, a deviation between WW ext and WW eff was already expected. The size of these valleys depends on the dilution, which in welding is defined as the percentage of base metal that blends with the added material (wire) composition. In this condition, the effective width values would only approximate the external width value when the dilution between the layers is such as to significantly reduce the formation of valleys in the side surface of the walls. Thus, in cases where dilution is low, more profound valleys tend to form, always leading to less effective widths than the external ones. As expected, Fig. 6b shows that the layer height decreases as TS is increased for the same WFS, since this variation reduces the amount of material deposited per unit of length. Nevertheless, when WFS is increased for a same TS, the opposite behaviour happens, leading to higher layer heights. Surface waviness (SW) is another geometrical feature that can be adopted as a function of WFS and TS in a working envelope approach. Figure 7a presents the average values for SW and its fitting curves, whereas Fig. 7b shows a representation of the same data in the form of a 2D response surface made via a commercially available statistical analysis software package. A prediction equation (Eq. 1) was determined by the software to visually express the SW data as a contour plot. It is worth mentioning that the SW assessed One must remember that an operating range of travel speed was established to avoid such irregularities. By analyzing Fig. 7a, one can see that when the same level of WFS is considered, surface waviness tends to reduce with increasing TS. On the other hand, this behaviour is not straightforward for faster WFS, like 5 m/min. Since the differences obtained between a given test and its adjacent ones are minimal (only about 0.1 mm), the trends carry some degree of uncertainty, justifying the difference in waviness trends at the three WFS levels. However, in general, as it can be seen through the contour lines in Fig. 7b, greater waviness occurs for lower TS and higher WFS (larger molten pool volume and higher arc pressure), and a smoother surface for the other way around (smaller molten pool volume and low arc pressure), respecting the limits of TS for high and low WFS. Validation of the reference working envelope Three walls were deposited with different combinations of WFS target and TS aiming at a target effective width of 4.5 mm (arbitrarily chosen) to validate the reference working envelope. The values of WFS and TS (in the second and third columns in Table 5) were defined by interpolation within the reference working envelope (Fig. 6a) from the target effective wall width. As evidenced, none of the combinations coincided with those used to build the working (Table 2), a principle of validation approaches. All other parameters were kept the same as in the construction of the reference working envelope. The remaining columns of Table 5 show the predicted and measured external wall width (WW ext ), effective wall width (WW eff ), layer height (LH) and surface waviness (SW). The predicted WW ext and LH were also reached by interpolation in the envelopes of Fig. 6a, b, respectively. The predicted SW was obtained by using Eq. (1). Table 6, in turn, contains the deviations between predicted and measured values. Within the geometric features assessed, WW ext had the highest deviations between the measured and predicted values, varying at around ± 0.3 mm. All the other features presented deviations between ± 0.1 mm. These low deviations indicate the robustness of the working envelope approach (the fact that the deviations present no tendencies concerning being greater or less than the predicted values also suggests statistical reliability). As shown in Fig. 8, the arc energies per unit of length calculated for Table 5 trials (in which different WFS and TS combinations reaching the same target effective width) were very similar (considering the standard deviations amongst layers). The wall geometries were practically the same (suggesting that the heat flow through the wall was similar). This behaviour was expected, since heat flux is the most important governing parameter to define the geometrical wall features. Accordingly, the thermal cycles experienced by the layers probably followed the same tendency (all facts indicating a strong correlation between arc energy and heat input). Consequently, no significant differences in microstructures were observed in the samples built in the validation trials, as shown in Fig. 9. However, it is well known that microstructure is governed by arc energy and thermal cycle and by chemical composition. In this sense, chemical composition analysis was carried out with a fluorescence x-ray spectrometer (XRF) to assess the possible influence of the deposition parameters over the burning losses of elements. In phase with the previous results, Table 7 suggests no significant variations (or trends) when chemical compositions were quantified over the extreme WFS and TS conditions. One can also assume that no chemical variation occurred between the walls from the validation trials. Naturally, microstructure changes were observed between the top layer and the remaining layers that underwent thermal retreatment from multiple thermal cycles. This behaviour was also observed by Aldalur et al. [20] and Kozamernik et al. [21] when using WAAM with the same ER70S-6 wire. To illustrate this behaviour, Fig. 10 presents macro and micrographs for the intermediate condition within the validation trials. As seen, no imperfections are observed. It can be noticed that the last deposited layer, which corresponds to a region not subject to reheat from the following layers (region 1), presents large columnar grains composed majorly by grain boundary ferrite (PF(G)) and acicular ferrite (AF). Region 2 illustrates that the fusion line is not easily perceived by optical microscope. Typical microconstituents of primary solidification zones are presented, such as grain boundary ferrite (PF(G)), acicular (AF), side plate ferrite (FS(A)) and some veins of polygonal ferrite (PF). Both the middle (region 3) and bottom regions of the wall (region 4) do not present major differences in microstructure between themselves and contain mainly polygonal ferrite. It can be said that these regions experience a similar thermal history based on the microstructure. However, it is worth noting that the microstructures of regions 1 and 2 occur only at the top surface of a thin wall and can be easily machined Table 6 Deviations between predicted and measured geometric features shown in Table 5 Wire out if it is the case. The principal volume of a wall will be composed of microstructures illustrated in regions 3 and 4, as long as the thermal management during the building keeps the same interlayer temperature. Notwithstanding, regions 1 and 2 turn to be relevant in short walls concerning mechanical functionality. This is the reason to present this microstructural feature in the current study. Figure 11 presents the microhardness profiles from the validation trials. There is no significant variation of the mean hardness when the three parametric conditions are compared (coherent with the microstructural features), as presented in Fig. 12. The average microhardness at the multi heat-treated regions is around 185 HV, within a narrow range of between 170 and 200 HV. The top region, in turn, shows a broader variation (of between 170 and 245 HV) as a consequence of the different ferrite morphologies typical of primary solidification. As evidenced in Fig. 13, the higher microhardness values in the top region are associated with regions rich in acicular ferrite. Exploring the potential of the working envelope approach for parameter transferability Three walls were deposited to assess the transferability of parameters from an existing working envelope when changing the feedstock, now using a high strength low alloy steel wire (AWS ER90S-B3). The same parameter combinations (columns 2 and 3 of Table 5) used for the reference envelope validation were replicated. Consequently, a similar estimated wall width as the walls for the reference envelope validation is expected at the outset. Thus, the following discussions are based on the comparison between the walls built for both transferability and validation of the reference envelope. In this context, Fig. 14 presents the appearance of the walls, in which no significant changes in the surface aspect are noticed when contrasting walls made with the same wire. However, more regular surfaces were achieved with the AWS ER90S-B3 wire when compared to AWS ER70S-3. It is also noticeable in this figure that WFS matches the target values for both wires, indicating that the walls were deposited with the same set of parameters. Table 8 presents the predicted and measured geometric features, namely, external wall width (WW ext ), effective wall width (WW eff ), layer height (LH) and surface waviness (SW). Table 9 presents the deviations (measured values minus the predicted ones). It can be verified that the external wall widths were always narrower and heights taller than expected, a difference not so significant when effective width is taken into account. Surface waviness was always less than predicted, reaching deviations of up to -0.3 mm, quantifying the best surface finish observed for the builds with the AWS ER90S-B3 wire, shown in Fig. 14. Since there is a difference in chemical composition between both materials, divergences could be expected. To better understand how the physical properties of the molten pool could have changed between the wires, Table 10 presents the chemical compositions for both materials. These correspond to the mean values shown in Table 7 for the AWS ER70S-6 wire and the mean values obtained for the walls deposited with the AWS ER90S-B3 wire, which will be presented later (Table 11). Deng et al. [22] proposed Eq. (2) to estimate the dynamic viscosity (η) of liquid steels as a function of the chemical elements, where T corresponds to the temperature within an optimum range between 1463 and 1723 K. The authors verified that the dynamic viscosity increases as the Si and Ti contents are increased, whilst an inverse effect is caused by increasing contents of Mn, P and S. Thus, although AWS ER90S-B3 has a lower Si content compared to AWS ER70S-6 (0.40 wt% on average), which would lead to a decrease in viscosity, the lower Mn content (0.72 wt%) would overcome this effect and cause the AWS ER90S-B3 molten pool to have a higher viscosity, resulting in higher resistance to movement induced by arc pressure and, as a consequence, the resultant bead would be narrower (wall width) and taller (layer height). It is worth mentioning that, as shown in the referred equation, Mn content has a higher coefficient when compared to Si content, indicating that Mn has a stronger effect over molten metal viscosity. Besides that, the existence of some surface-active agent (S, O, Se and Te), even in small amounts, could change wettability and lead to changes in molten pool geometry. For instance, Keene et al. [23] observed large variations in surface tension between 316 stainless steel grades with a difference of 139 ppm (0.0139%) of S. Finally, both viscosity and surface tension are highly dependent on temperature, so possible changes in thermal diffusivity between the materials may also affect the wall geometry. To discuss the effect of the feedstock on arc physic aspects, Fig. 15 presents the average values from the main process parameters monitoring (mean and RMS current, mean and RMS voltage). Slightly higher mean and RMS current can be seen in the walls built with AWS ER90S-B3 (Fig. 15a, b), although the same WFS and TS had been set. This means that a higher current is needed to melt the AWS ER90S-B3 wire at the same melting rate. To investigate this behaviour, mean values of arcing time (t arc ) and short-circuiting time (t sc ) were calculated through the electric signal data from five layers, using a home-developed software (CURTOWELD), registered by Vilarinho and Araújo [24]. The synergy line of CMT imposed on these materials 50% of arcing and short-circuiting times, but with different durations. For the AWS ER70S-6 wire, t arc and t sc presented the same value equal to 6.3 ms, while with AWS ER90S-B3, they were equal to 6.0 ms. As a result, the short-circuiting frequency with ER90S-B3 (82.9 Hz) was slightly higher than with ER70S-6 (79.8 Hz). A slight increasing shifting between current and voltage waveforms from both wires can be seen in Fig. 16, due to the difference in short-circuiting frequency. As noticed in Fig. 16, the currents at arcing and short-circuiting times are similar to the two wires. Short-circuit frequency by itself does not justify the higher current levels for AWS ER90S-B3 to maintain the same melting rate (same WFS). For this to happen, it would be necessary that the t arc /t sc ratio be greater for the low alloy steel wire, which does not occur. However, this current related behaviour can be discussed with the aid of the equation for melting rate in short-circuiting metal transfer [25]. In this equation, MR is the melting rate, α and β are constants that depend on electrode polarity, shielding gas composition, wire material, ρ is the electric resistivity of the wire, L is the electrified wire free extension, S is the cross-section area of the wire, I m is the mean current, I rms the root mean square of current, t arc is the arcing time and t sc is the short-circuiting time. Knowing that the same melting rate was achieved for a given combination of WFS and TS (Fig. 14b) and considering that the arc length was the same for both wires (intrinsic due to the use of the same synergic line), the higher current with AWS ER90S-B3 can be justified by the following hypothesis: a) cross-section area (S) wire is larger; b) a lower electrical resistivity (ρ); c) α and β values become lower. The first hypothesis was confirmed by measuring the wire diameters with a micrometer (seven measurements in each wire) and the results showed that the diameters were 1.17 ± 0.01 mm and 1.15 ± 0.0 mm, respectively, for AWS ER90S-B3 and AWS ER70S-6 (leading to areas of 4.30 e 4.15 mm 2 ). One evidence that could lead to the conclusion that the low alloy steel wire attains lower electrical resistivity (ρ) would come from the marginally lower mean and RMS voltages (Fig. 15c, d) when this wire was used. The third hypothesis, related to the constant α e β, was not possible to be assessed in this work. Whatever the reason or combination, it is demonstrated that the two wires had different arc physics properties, not only different chemical compositions. Hence, their use to demonstrate the potential of parameter transferability of the approach working envelope is assured. Closing the analysis of the electrical signals, Fig. 17 demonstrates that, similar to the validation trials (Fig. 8), no significant difference amongst the arc energies per unit of length is observed amongst the 3 walls employed to achieve the same effective wall width (4.5 mm) with AWS ER90S-B3 as feedstock (considering the standard deviations). However, other combinations of WFS and TS might deliver different results. Therefore, it was not surprising that the wall geometries were practically the same, as seen in Tables 8 and 9. Furthermore, according to Table 11, no variation in the deposited chemical composition was observed either. As a consequence of the similarity between the results of arc energy, geometry and chemical composition, Fig. 18. The same analysis applied to the reference working envelope towards the variation of microstructure and hardness along the wall building direction was replicated to the trials with the AWS ER90S-B3 wire. The condition with WFS = 3.8 m/min and TS = 38.5 cm/min is taken as an example and macro and micrographs from this trial are shown in Fig. 19. As seen, no imperfections are observed. Contrasting with the walls made of C-Mn steel (AWS ER70S-6), no significant changes in macro and microstructure are noticed even between the top layer and the remaining layers. More significant formation of tempered martensite is possible in the transition between the last two layers deposited (region 2), the middle of the wall (region 3) and in the bottom (region 4) when compared to the top layer (region 1). This is reasonable since these regions are subject to reheating by deposition the following layers and could be tempered. However, to correctly affirm the above, a more thorough microstructure investigation should be carried out, which is out of this work context. In this case, it can be affirmed only that microstructure is composed mainly of martensite, tempered martensite and bainite, which is in accordance with the findings of Dirisu et al. [19] and Sharma and Shahi [26] for AWS ER90S-B3. Microhardness values ranged within 270 and 420 HV for the AWS ER90S-B3 walls, as can be seen in Fig. 20. A cyclic behaviour, indicated by red arrows, was identified. Depending on the peak temperatures and cooling rates of the multiple thermal cycles experienced, each region may present a higher or lower formation of martensite/bainite and lead to the observed result. The average microhardness values were around 335 HV, as shown in Fig. 21 with the respective standard deviations. As a whole, there are differences in the metallurgical architecture between the walls for the envelope validation and the walls for evaluating the parameter transportability because the feedstocks are different. But the transportability of the parameters showed to be feasible. The use of the working envelope approach for parameter optimization To investigate the complementary objective as to the possibility of optimizing parameter selection for a same target width, Fig. 22 presents the reference working envelope for effective wall width considering arc energy per unit of length (E), surface waviness (SW) and active deposition time (t ad ) as responses. It must be highlighted that active deposition times were determined considering a wall with 140 mm in length, with a total height of 40 mm. First, the number of necessary layers was found by dividing 40 mm by the layer heights achieved for each WFS and TS combination (Table 4). Next, the number of layers for each condition was multiplied by the wall length (140 mm), considering the distances travelled. These distances were finally divided by the travel speeds, resulting in active deposition times (not considering dwell times). Second-order models, defined in Eq. (4), were used to plot the contour lines. Aiming at improving the variability proportion that can be quantified for each model (R 2 ), the significance level of each term of the model was evaluated through ANOVA and the less significant terms (p-values>0.05) were discarded. Equations (5), (6) and (7) present the models used to develop the surfaces in Fig. 22 and their respective R 2 and adjusted R 2 . According to Montgomery [27], adjusted R 2 corresponds to a variation of R 2 that is adjusted to the model's size, that is, the number of factors. R 2 higher than 0.95 means that 95% of the data variability can be explained by the models. Figure 22 also displays the respective observed-bypredicted charts on the right side of each contour plots. These charts are helpful to detect misspecifications in the structural model. Ideally, values should lie roughly along a 45-degree line. Predictions that are outside the interval are denoted as outliers. A high proportion of Table 9 Deviations between predicted and measured geometric features shown in Table 8 Wire where f(x,y) is the predicted or expected value, i.e. the regression function of the independent variables "x" and "y" (which in this case correspond to travel speed and effective wall width); "A" corresponds to the intercept (predicted value when all of the independent variables are equal to zero); and "B", "C", "D", "E" and "F" are the estimated regression coefficients. (4) f (x, y) = A + Bx + Cx 2 + Dy + Ey 2 + Fxy (Fig. 17) were not enough to significantly change microstructure or microhardness profiles for a target width of 4.5 mm. Although this difference can become larger for higher widths, it is still unlikely that they are sufficient to lead to significant changes in microstructure and mechanical properties. Hence, for the reference working envelope, or a potential working envelope (AWS ER70S-6) considering the transferability to AWS ER90S-B3, it can be stated that, for a given target width, variations in WFS and TS do not significantly affect microstructure and microhardness. Figure 22b shows that employing lower values of WFS and TS results in smaller SW and, consequently, better surface finishing for a same effective wall width. Considering Eq. (6) and taking, for instance, 4.0 mm as WW eff and values of TS equal to 40.0 and 57.0 cm/min (values close to the envelope boundaries), the resulting SW are 0.5 and 0.8 mm, respectively, resulting in a difference (ΔSW) of 0.3 mm. If a WW eff of 6.5 mm is taken, with TS values of 16.0 and 22.0 cm/min (again close to the envelope boundaries) the resultant SW will be 0.9 and 1.0 mm, respectively, resulting in a ΔSW of 0.1 mm. This means that for larger widths, the surface finish deteriorates. However, it is less affected by the WFS/TS combination. It is important to draw attention to the fact that the low adjusted R 2 of Eq. (6) may become feasible if the correspondent contour plot represents this quantity. The observed-by-predicted chart for SW shows that the values lie reasonably along a 45-degree line and that the deviations are lower than 15%. Moreover, the distribution of the observations is symmetrical around the corresponding predicted values. All of this gives statistical confidence to the estimator, even with a relatively low adjusted correlation coefficient. Finally, Fig. 22c indicates that by employing higher values of WFS and TS, for a given target width, active deposition time (t ad ) becomes shorter. Elapse times may be in the opposite direction at first sight. However, the reader must remember that for this material and power source, arc energy presents low changes for a given wall width (elapse time is roughly proportional to arc energy). Changes in the desired width also influence the deposition times found for different WFS and TS combinations. Based on Eq. (7), taking WW eff as 4.0 mm and TS at 40.0 and 57.0 cm/min, the resulting t ad are 7.4 and 4.3 minutes, respectively, resulting in a difference (Δt ad ) of 3.1 min. When the WW eff of 6.5 mm is taken, with TS varying from 16.0 to 22.0 cm/min, the resulting t ad are 11.5 and 8.5 min, respectively, resulting in a Δt ad of 3.0 min. Although the t ad does not take dwell times (necessary to reach the interlayer temperatures of 30 °C) into In summary, the approach applied in this work to optimize parameter selection allowed not only for visualization of the effects of each parameter over operational features desired for a same effective width, but also a quantification through equations. Computational optimization routines could be used to generate the optimum combination between TS and WFS for a desired wall width using arc energy, surface waviness, active deposition time and others, as an objective function or as restrictors. For instance, one could wish to find TS and WFS to build a wall with a determined width to provide at the same time the shortest active time, with a minimum surface waviness and with arc energy higher than a given value. Nevertheless, it must be highlighted that the interactions here discussed cannot be stated straightforward for working envelopes using other processes or materials. However, they worked well between the two materials evaluated. Conclusions This work was motivated by making the working envelope approach for WAAM parameterization more cost-effective and functional. The objective of this work was to verify the possibility of parameterizing thin walls based on a pre-existing working envelope, followed by investigating whether the approach can be used to optimize parameter selection. Based on these objectives and results, it can be concluded that: • It is possible to select parameters for thin walls deposited through WAAM with a lower number of experiments, using a pre-existent working envelope for another ordinary material (parameter transferability), implying lower time and resource consumption before the deposition of a final component. However, it is essential to mention that deviations between predicted and measure values for different materials are highly dependent on the feedstock compatibility in terms of the physical-chemical properties of both materials. Thus, not every parameterization with other materials (different classes), using this approach will be necessarily successful without some degree of adaptation. • The working envelope combined with contour lines also demonstrated to be an easy means of visualizing the effect of the primary process variables (travel speed and wire feed speed). Optimization for different objective functions and restrictors can be implemented in the working envelope approach, either manually or using computational resources.	2022-11-21T14:22:52.270Z	2021-11-12T00:00:00.000	{ "year": 2021, "sha1": "2f78053ba7c2c8c700eaa2d64b6d298394c3dca7", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/s00170-021-08326-2.pdf", "oa_status": "HYBRID", "pdf_src": "SpringerNature", "pdf_hash": "2f78053ba7c2c8c700eaa2d64b6d298394c3dca7", "s2fieldsofstudy": [ "Engineering", "Materials Science" ], "extfieldsofstudy": [] }
253241904	pes2o/s2orc	v3-fos-license	Dispersal Kernel Type Highly Influences Projected Relationships for Plant Disease Epidemic Severity When Outbreak and At-Risk Populations Differ in Susceptibility In silico study of biologically invading organisms provide a means to evaluate the complex and potentially cryptic factors that can influence invasion success in scenarios where empirical studies would be difficult, if not impossible, to conduct. I used a disease event simulation program to evaluate whether the two most frequently used types of plant pathogen dispersal kernels for epidemiological projections would provide complementary or divergent projections of epidemic severity when the hosts in a disease outbreak differed from the hosts in the at-risk population in the degree of susceptibility. Exponential dispersal kernel simulations of wheat stripe rust (Puccinia striiformis var trittici) predicted a relatively strong and dominant influence of the at-risk population on the end epidemic severity regardless of outbreak disease levels. Simulations using a modified power law dispersal kernel gave projections that varied depending on the amount of disease in the outbreak and some interactions were counter-intuitive and opposite of the exponential dispersal kernel projections. Although relatively straightforward, the disease spread simulations in the present study strongly suggest that a more biologically accurate dispersal kernel generates complexity that would not be revealed by an exponential dispersal gradient and that selecting a less accurate dispersal kernel may obscure important interactions during biological invasions. Introduction Complex systems are difficult to study empirically, but its components can be understood or at least statistically described in a way that the information can be used to create models to project responses under scenarios that may be impossible to create experimentally [1][2][3]. For invasion biology, models are an important tool for projecting the spatio-temporal patterns of a biological invasion, and they can also facilitate investigations into difficult to study factors and how they may suppress or encourage organism invasion. The insights gained from carefully constructed models containing well-established ties to biologically realistic mechanisms can be crucial for implementing mitigation strategies to control the invading organism [4,5]. While in silico studies are an obvious departure from on-the-ground empirical study and require simplifying assumptions, they are an important method to understand and project the impacts and spatio-temporal patterns associated with biological invasions. A wait and see strategy for the empirical study of a biological invasion is simply not pro-active enough given the world-wide loss of biodiversity, human life, and ecosystem changes that are now the text-book outcomes of uncontrolled biological invasions. To create a potentially useful statistical model of organism invasion, the stages of the invading organism in terms of colonization, reproduction, and dispersal need to be integrated and preferably run in a spatially explicit, virtual landscape [6]. Demographic rates (e.g., vital rates) and colonization probability can be measured through observation and/or estimated through direct empirical study, manipulative experimentation, and/or Life 2022, 12, 1727 2 of 11 in combination with in silico methods that use sensitivity analyses and pattern-oriented modeling [2,6]. One of the more difficult, but critically important, aspects of biological invasion models to measure and/or estimate is the dispersal kernel. The dispersal kernel is a mathematical function that is used to statistically describe how an organism disperses through a landscape over time. Dispersal kernels are notoriously difficult to measure and to accurately parameterize for organisms that are prone to rare, long-distance dispersal events. The challenge to represent the rarer successful long-distance dispersal events are that the successful events are sparse and embedded within a large expanse of absences and this is type of data is information poor compared to the area near an invasion source which contains a relatively large number of successful dispersal events over shorter distances. Because these long-distance events are rare, they can be easily underestimated by a dispersal kernel but be biologically meaningful for the patterns of invasion spread [7]. Disease epidemics are considered a form of biological invasion [8] and they share similar factors that influence epidemic severity as well as control philosophies [9][10][11][12][13]. For plant pathogens, there are two primary types of dispersal kernels that have been used to project the spread of aerially vectored plant diseases (primarily fungal diseases). One family of dispersal kernels are those with functions that are exponentially bound (e.g., exponential, double exponential) and the other family consists of those functions that are not bound by an exponential (e.g., modified power law, modified Pareto distribution) [7,14]. Exponential functions have longer distributional tails (more kurtotic) than a normal distribution and describe the decrease of inoculum/disease dispersed from a source over a greater expanse with rapidly decreasing disease levels as the distance from the source increases. Exponential family dispersal kernels eventually terminate when either the fitted function to empirically collected data crosses the x-axis or the probability of occurrence reaches zero (for probability density functions). The non-exponentially bound dispersal kernels are comparatively more leptokurtic (fatter, thicker, or heavier tails) than the exponentially bounded functions, with the distribution's tails extending for a much greater distance at very low predicted probabilities. In comparison, these more kurtotic functions expand the small probability of long-distance dispersal events over a much greater distance than exponential kernels. It is a mathematically demonstrated outcome that if the amount of host is approximately continuous and homogenously distributed, and in a sufficiently sized area, that exponentially bound functions will produce disease invasion fronts that move through the host population with a constant rate following a short period of acceleration [14,15]. These exponentially bound dispersal kernel functions simplify to a diffusion rate, a constant rate of disease spread over space, and this property facilitates straightforward predictive diffusion-based epidemiological projections. However, there is also empirical evidence that wind vectored plant diseases are inadequately described by an exponential function (the function's tails are significantly truncated compared to the actual observed dispersal gradient) and that non-exponentially bound functions (dispersal kernels with much longer distribution tails) are biologically more appropriate [7,[16][17][18]. In contrast to the exponential family of dispersal kernels, the long-tailed, non-exponentially bound dispersal kernels produce disease invasions with fronts that appear to always increase in velocity over space until host and/or space become limiting, and therefore cannot be represented by a rate constant, even as a simplification [14,16,18]. Provided the same raw data which were modeled under the same environmental (and host) conditions, these two dispersal kernel types not only generate different rates of organism spread but they also predict markedly different patterns of disease abundance with respect to its source [14][15][16][17]. The issue of disease susceptibility, especially as it pertains to understanding and projecting the spread of disease, is an important topic given that vaccinations are expected to generate specific outcomes in the at-risk population and disease resistance bred into plants should suppress disease. However, this issue is not straightforward to study empirically, as between field borders can differ in cultivar composition, fields may be intercropped, cultivar mixtures can be planted, and even alternating rows of different cultivars and fungicide treatments (a cost saving technique that lowers fungicide application rates) are not uncommon grower practices. For such scenarios, it reasonable to ask whether there is a suppressive or facilitative influence (and whether this impact may be predictable) on subsequent epidemic severity when disease disperses from the outbreak into an at-risk population where host resistance is either greater or lower than that of the outbreak host population. For the purposes of this manuscript, I consider the outbreak to be the area (and its host plants) that the initial disease generation occupies and the at-risk population to be all hosts outside of the outbreak. It is possible that the answer to this question could be purely demographic in nature -simply that the reduction or increase in relative reproductive rates is the primary determinant of later disease severity in an at-risk population. However, the shape and degree of dispersal kernel kurtosis can generate a strong impact on the subsequent patterns of spread and the spatial patterns of disease intensification from an outbreak as it spreads into the at-risk population [14,16,[19][20][21]. I used a series of in silico experiments to understand whether the dispersal kernel type, exponentially bound or non-exponentially bound, substantively influences the patterns of disease projections when disease transitions between outbreak and at-risk host populations that differ in disease susceptibility. I focused on wheat stripe rust, an economically important, world-wide, disease of wheat caused by the fungus Puccinia striiformis var tritici (hereafter Pst), a well-studied and relatively well characterized plant pathosystem from an epidemiological perspective. In particular, I was interested in evaluating whether one or both dispersal kernel types (exponentially bound or non-exponentially bound) could yield relationships that are consistently predictable over a range of disease outbreak levels and whether those projections are similar enough to suggest that a simplified approximation could be made about the potential interactions. For example, it is possible that the overall difference in susceptibility between the outbreak and at-risk populations proportionally increases or decreases the amount of disease in the at-risk population according to a predictable linear relationship. Wheat Stripe Rust Wheat stripe rust (WSR) is caused by the fungus Puccinia striiformis var tritici (Pst) and it is an obligate parasite of its host plant (obligate plant pathogens require relatively healthy and vigorous hosts for disease to occur). WSR can be encountered wherever wheat is grown [22][23][24][25] and its alternative host plants appear to be Berberis spp. [26,27]. However, it is unlikely that Berberis spp. are necessary for WSR epidemics as Pst spores can overwinter in the soil and thatch when conditions are mostly above freezing [23,28]. Pst produces spores (~10 to 20 microns which appear to be somewhat environmentally resilient to temperature and some UV light exposure [29]) that are borne on uredinia in small aggregates referred to as pustules. Groups of pustules form lesions, which are presented linearly on the upper and lower leaf surfaces, and elongate over time parallel to wheat leaf veins, yielding the "striped" appearance of WSR. Spores are produced in large amounts, several hundred or more uridineospores/day per square millimeter of lesion [25], and R 0 (the basic reproductive number, the mean number of daughter infections arising from a single infection) can be very high (ranging from 35 to~800) depending on host availability and pre-existing disease levels [30,31]. Disease occurs as long as the wheat plant can physiologically support either new infections or the expansion of existing lesions. As WSR outbreaks intensify, the disease grows at an exponential rate [10], but successful dispersal events can occur over large distances even from relatively small outbreaks [32] and rarer long-distance events are known at continental scales [28,33]. Although wheat stripe rust can be theoretically well-managed through the appropriate timing of fungicide applications [34], WSR epidemics can cause massive damage on susceptible wheat cultivars [23,24,28,35]. Unfortunately, there is also recent evidence that some Pst lineages have evolved fungicide resistant mutations [36], which has caused considerable problems for the management other wheat fungal diseases on wheat such as Septoria leaf blotch (Zymoseptoria trittici) [37][38][39][40], eyespot (Oculimacula spp.) [41], and wheat blast (Magnaporthe oryzae) [42]. With the increasing incidence of fungicide resistant wheat plant diseases across the world, including Pst, control will probably be accomplished through the breeding of durable disease resistance [43]. This means that understanding how disease susceptibility may alter epidemic behavior is an important aspect to understand going forward. Wheat Stripe Rust Disease Spread Simulations I used an updated and highly modified version of the plant disease simulation program EPIMUL [44] to run the in silico projected disease spread experiments. EPIMUL is a spatially explicit, compartmental disease event simulator, which is parameterized to represent real space in a wheat field. Each compartment was filled with virtual host plants that were similar in density to production fields and previously published empirical studies of WSR spread. Plants within the compartment were assigned properties (e.g., density, disease carrying capacity, latent and infectious periods, disease reproduction rates, infection probability, outbreak or at-risk population) and effective disease spores were distributed across this landscape according to a specified dispersal kernel. The epidemiological variables used in the present model originated from published intensive field studies performed in western and central Oregon, USA (see below for simulation and parameter details). For this study, I used deterministic simulations as I was interested in the mean differences between scenarios rather than focusing on the variation within a single scenario and how that variation overlaps with a slightly different set of parameter values. In EPIMUL, stochasticity is built into the dispersal gradient as a Poisson resampling of the original dispersal gradient [10]. In previous simulations, the mean disease levels over space and time in each compartment from 100 stochastic simulations was nearly equivalent to one deterministic run in EPIMUL [10], so while there was information in variability to be gained from stochastic simulations this approach was not necessary given the goals of the present study. Compartment parameters for the simulations were consistent with previous WSR simulations [17,45] and updated with more accurate parameter values when supported by newer published data. The simulation field size was 800 × 800 compartments, with each compartment having dimensions of 1.52 m × 1.52 m (the width of a wheat planter) and each compartment had a carrying capacity of 200,000 infection sites, which is the average number of sites estimated from a standard wheat planting density over the life of the average wheat plant [17]. I used a latent and infectious period of 12 days, which is common for WSR outbreaks in the late spring and early summer when conditions are optimal for the disease in central Oregon. R 0 , the basic reproduction number [46], which is the mean number of daughter infections arising from a single mother infection, was set at 70 for the completely susceptible genotype and reduced proportionally with a decrease in susceptibility (an increase in disease resistance). This method is described below in a separate paragraph. The fully susceptible host R 0 = 70 is consistent with previous experiments featuring fully susceptible and partially susceptible wheat genotypes [45,47] and studies of WSR development [30] over a range of environmental conditions that were comparable to central Oregon. I used two different dispersal kernels to simulate WSR disease spread in the exact same virtual field arrangement to understand the influence of each dispersal kernel type on epidemic projections. The first gradient was the modified power law dispersal kernel reported by Farber et al. [32]. This is the most accurately and precisely described dispersal gradient for WSR available in the published literature. For the modified power law, the dispersal kernel was described by the formula y = a (x + c) −b where "a" was a value that adjusts the amount of disease produced at the source; b modified the steepness of the dispersal kernel; and the c value allowed for the power law dispersal kernel to have a non-zero value when x = 0 and also modified the kernel shape. For the modified power law simulations, the values of each variable were: a = 425, b = 2.28, c = 0.23. The exponential function was calculated from the original data used by Farber et al. [32] (which was originally and appropriately best-fit to the power law kernel above) and an exponential model was forced on the Farber et al. [32] raw data with the method traditionally used by plant pathologists to fit disease gradients to an exponential kernel [48]. The exponential dispersal kernel was described by the following formula: y = a exp (−bx); for this study a = 19.2, b = 0.1903. WSR infections were dispersed equally (radially) from the source using the downwind dispersal gradient reported by Farber et al. [32]. I also evaluated the potential influence of the amount of disease in the outbreak on the projections. Disease levels in the outbreak can have a strong and dominant impact on the severity of the subsequent WSR epidemic in the at-risk population, in field experiments [45,47] and in simulations [10]. It is possible there were dispersal kernel × outbreak disease level interactions that influence epidemic severity when host populations differ in disease susceptibility. The outbreak levels of disease in my simulations were set at 0.05%, 1.0%, and 5.0% of the total sites available (disease carrying capacity), and these values span the range of biologically reasonable outbreak levels (0.05% and 1.0%) and exceptionally high outbreak levels (5%). I set up two virtual landscapes that were used with both dispersal kernels and each disease outbreak level to project the interactions of epidemic severity given the differences in host disease susceptibility in a standardized landscape. Both fields contained an outbreak (focus) that was one compartment (1.52 m × 1.52 m) in the center of an 800 × 800 compartment landscape. All compartments other than the outbreak represent the at-risk host population. In one scenario, the outbreak compartment was always 100% susceptible but the at-risk population host susceptibility varied in increments of 10% (from 100% to 10%). A susceptibility of 0 would not generate disease in the model as these hosts are completely resistant and useless in the present study. In the second scenario, the at-risk population was always 100% susceptible but the outbreak varied in susceptibility by increments of 10% ( Figure 1). To compare the relative effect of the transition from populations of host that differed in susceptibilities, an internal control, I simulated disease spread in monocultures for the same increments of susceptibility (e.g., 10% focus to 10% at-risk, 50% focus to 50% at-risk, 100% focus to 100% at-risk). gradient for WSR available in the published literature. For the modified power law, the dispersal kernel was described by the formula y = a (x + c) −b where "a" was a value that adjusts the amount of disease produced at the source; b modified the steepness of the dispersal kernel; and the c value allowed for the power law dispersal kernel to have a nonzero value when x = 0 and also modified the kernel shape. For the modified power law simulations, the values of each variable were: a = 425, b = 2.28, c = 0.23. The exponential function was calculated from the original data used by Farber et al. [32] (which was originally and appropriately best-fit to the power law kernel above) and an exponential model was forced on the Farber et al. [32] raw data with the method traditionally used by plant pathologists to fit disease gradients to an exponential kernel [48] (K. Sackett pers. com.). The exponential dispersal kernel was described by the following formula: y = a exp (−bx); for this study a = 19.2, b = 0.1903. WSR infections were dispersed equally (radially) from the source using the downwind dispersal gradient reported by Farber et al. [32]. I also evaluated the potential influence of the amount of disease in the outbreak on the projections. Disease levels in the outbreak can have a strong and dominant impact on the severity of the subsequent WSR epidemic in the at-risk population, in field experiments [45,47] and in simulations [10]. It is possible there were dispersal kernel × outbreak disease level interactions that influence epidemic severity when host populations differ in disease susceptibility. The outbreak levels of disease in my simulations were set at 0.05%, 1.0%, and 5.0% of the total sites available (disease carrying capacity), and these values span the range of biologically reasonable outbreak levels (0.05% and 1.0%) and exceptionally high outbreak levels (5%). I set up two virtual landscapes that were used with both dispersal kernels and each disease outbreak level to project the interactions of epidemic severity given the differences in host disease susceptibility in a standardized landscape. Both fields contained an outbreak (focus) that was one compartment (1.52 m × 1.52 m) in the center of an 800 × 800 compartment landscape. All compartments other than the outbreak represent the at-risk host population. In one scenario, the outbreak compartment was always 100% susceptible but the at-risk population host susceptibility varied in increments of 10% (from 100% to 10%). A susceptibility of 0 would not generate disease in the model as these hosts are completely resistant and useless in the present study. In the second scenario, the at-risk population was always 100% susceptible but the outbreak varied in susceptibility by increments of 10% ( Figure 1). To compare the relative effect of the transition from populations of host that differed in susceptibilities, an internal control, I simulated disease spread in monocultures for the same increments of susceptibility (e.g., 10% focus to 10% at-risk, 50% focus to 50% at-risk, 100% focus to 100% at-risk). Schematic representation of two different landscape simulation scenarios where the outbreak (focus) and the at-risk population differed in the degree of quantitative resistance (susceptibility) by increments of 10% through the proportional reduction of R 0 (see methods below). The left field depicts the scenario where the outbreak (focus) is comprised of a 100% WSR susceptible genotype and the at-risk population (the remainder of the field) decreases in the degree of susceptibility by increments of 10%. The right field depicts the opposite scenario where the focus is comprised of host plants that are variably susceptible and the at-risk population is 100% susceptible. Note that the focus and the at-risk field is not to the scale of the simulations. In the simulations the focus is considerably smaller relative to the at-risk field size. To model the differences in host susceptibility within the two landscape scenarios, I proportionally decreased R 0 in 10% increments and assigned the desired levels of susceptibility to the outbreak and at-risk compartments (e.g., 100% susceptible hosts had an R 0 = 70, 10% susceptible hosts have an R 0 = 7). I held the infection probability the same for the dispersed effective spores which, in combination with a decreased R 0 , reduced their capacity for disease production if infected. Although, this approach is overly simplistic, as biologically resistance can arise from different mechanisms (e.g., reduced infection probability, reduced virulence, smaller lesions, lower sporulation rates), proportionally reducing the R 0 is a straightforward method to represent hypothetical quantitative resistance from any mechanism and evaluate the resultant patterns of epidemic progression. To index the relative amount of disease that accumulated in the at-risk population from the outbreak after five disease generations (60 days), I calculated the area under the disease gradient (AUDG) for a 1 × 301 compartment area extending from the outbreak in a straight line (Figure 1). I subtracted the amount of disease in the outbreak compartment to arrive at an end epidemic AUDG value for the at-risk population. Calculating the amount of disease along a transect in the simulations mimics empirical studies of plant disease spread that sample disease at points along a straight line from the source [16,17,45,47,48]. I plotted the AUDG values for all different combinations of outbreak disease levels, monocultures, and disease accumulated in the at-risk populations and grouped the simulations by the two different landscape scenarios for comparison. Results There were general patterns of disease increase that were consistent regardless of the dispersal model. Overall, the AUDG values (an index of relative epidemic severity) was predictably greater when the amount of disease in the outbreak was greater (Figure 2). Additionally, when the outbreak and at-risk populations were both at 100% susceptibility, the projected amount of disease was the greatest observed, and when either the outbreak or at-risk population was comprised of host plants with 10% susceptibility, epidemic severity was the lowest observed ( Figure 2). However, projections from the two dispersal kernel types yielded differently shaped responses in the amount of disease accumulated after 5 generations. The exponential dispersal kernel projected relatively consistent epidemic responses when disease developed from an outbreak and intensified over time in at-risk population which differed from the outbreak in the degree of disease susceptibility ( Figure 2D-F). When the at-risk population was 100% susceptible, the at-risk population susceptibility exerted a dominant influence on the amount of disease that accumulated over time in the atrisk population, regardless of host susceptibility in the outbreak ( Figure 2D-F orange lines). When the outbreak was 100% susceptible, the at-risk population degree of susceptibility also strongly influenced the amount of disease that accumulated in the at-risk population ( Figure 2D-F blue lines). For both landscape scenarios, the projected relationships were approximately linear at the lower outbreak disease levels (0.05% and 1%), suggesting that host susceptibility of the at-risk population drives epidemic severity in a potentially straightforward and predictable manner. Only at the greatest outbreak disease level (5%), did the projected relationships become more curvilinear ( Figure 2F), but the influence of the at-risk host population susceptibility on the end epidemic severity was consistent with lower outbreak disease levels. There was no consistent pattern of end epidemic severity when disease was dispersed with the modified power law (Figure 2A-C). For each outbreak disease level, the influence of either the outbreak or the at-risk population's level of susceptibility appeared to generate different projections of end epidemic severity (Figure 2A-C). In all circumstances, including the monocultures, the projected epidemic severity relationships did not appear to behave in any obvious generalizable manner and different outbreak disease levels projected different relationships. For example, at the lowest outbreak disease level, 0.05%, the projected epidemic severities were strongly curvilinear while at the greatest outbreak disease level Life 2022, 12, 1727 7 of 11 (5% disease) the projected relationships were more linear than curvilinear compared with lower outbreak disease levels. Unlike the exponential dispersal kernel simulations, the outbreak could either strongly influence the epidemic outcome (Figure 2A), be roughly equivalent in its influence to that of the at-risk population ( Figure 2B), or be slightly suppressed by the level of susceptibility of the at-risk population ( Figure 2C). persed with the modified power law (Figure 2A-C). For each outbreak disease level, the influence of either the outbreak or the at-risk population's level of susceptibility appeared to generate different projections of end epidemic severity (Figure 2A-C). In all circumstances, including the monocultures, the projected epidemic severity relationships did not appear to behave in any obvious generalizable manner and different outbreak disease levels projected different relationships. For example, at the lowest outbreak disease level, 0.05%, the projected epidemic severities were strongly curvilinear while at the greatest outbreak disease level (5% disease) the projected relationships were more linear than curvilinear compared with lower outbreak disease levels. Unlike the exponential dispersal kernel simulations, the outbreak could either strongly influence the epidemic outcome (Figure 2A), be roughly equivalent in its influence to that of the at-risk population ( Figure 2B), or be slightly suppressed by the level of susceptibility of the at-risk population ( Figure 2C). Figure 2. (A) (0.05% outbreak disease levels), (B) (1% outbreak disease levels), (C) (5% outbreak disease levels) are the trends projected from AUDG (area under the disease gradient) values generated from the two field scenario simulations where the outbreak and at-risk populations varied in their relative degree of susceptibility (quantitative resistance to WSR) using the modified power law dispersal kernel. (D) (0.05% outbreak disease levels), (E) (1% outbreak disease levels), (F) (5% outbreak disease levels) are the same projections and disease summary statistics that were generated by the simulations using the exponential dispersal kernel (note that the susceptibility is the same, but it is presented differently on the x-axis, with (10) to draw attention that these figures were generated from the exponential dispersal kernel). Gray dots and trend lines represent simulation results from monocultures (e.g., 10% focus to 10% at-risk, to 100% focus to 100% at-risk), the blue dots and trend lines represent the scenario where the focus is 100% susceptible but the at-risk population has variable susceptibility, and the orange dots and trend line represents the scenario where the focus varies in susceptibility but the at-risk population is 100% susceptible. , (E) (1% outbreak disease levels), (F) (5% outbreak disease levels) are the same projections and disease summary statistics that were generated by the simulations using the exponential dispersal kernel (note that the susceptibility is the same, but it is presented differently on the x-axis, with (10) to draw attention that these figures were generated from the exponential dispersal kernel). Gray dots and trend lines represent simulation results from monocultures (e.g., 10% focus to 10% at-risk, to 100% focus to 100% at-risk), the blue dots and trend lines represent the scenario where the focus is 100% susceptible but the at-risk population has variable susceptibility, and the orange dots and trend line represents the scenario where the focus varies in susceptibility but the at-risk population is 100% susceptible. Discussion WSR disease spread simulations, which were calibrated against well-characterized demographic, epidemiological and dispersal parameters, yielded conflicting projections of how disease susceptibility may alter epidemic severity when the outbreak and at-risk host populations differ in their degree of resistance. These differences emerged as an interaction between the dispersal kernel type and the amount of disease that founded the outbreak. Host arrangement, including the virtual field size, compartment number and its dimensions, the location of the outbreak within the virtual field, the area from which disease was estimated, the latent and infectious periods, host density, generation time, infection probability and R 0 (for 100% susceptible genotypes) were standardized throughout the simulations. Disease was also dispersed in a radially symmetric manner from the focus (there was no asymmetric anisotropy; e.g., upwind, downwind, changing wind directions and magnitude) to keep the scenarios as straightforward as possible for comparison. Only the dispersal kernel, the amount of disease in the focus at the outbreak onset, and the susceptibility of host plants in specific compartments (through the proportional reduction of R 0 ) were modified. Despite this degree of standardization and constant conditions that are obvious departures from a "real life" WSR outbreak, simulations suggested that rule of thumb guidelines for predicting where, when and how much disease may be generated may be possible for organisms with exponential dispersal kernels but unrealistic for organisms characterized by long-distance dispersal. The penalty for over-simplification (a truncated dispersal gradient) was a suite of facile but potentially seriously misleading epidemic projections. These projections were attractive for suggesting a potential predictable pattern of disease spread, whereas attempting to reflect a more biological realistic scenario (through a well-fit dispersal kernel) gave a less intuitive assemblage of epidemic projections. There appeared to be an important tradeoff threshold between convenient interpretation and attempting to reasonably represent the biological reality of long-distance dispersal due specifically to the dispersal kernel. For the sake of disease management and projecting the impacts of having a mosaic landscape of hosts that differ in disease resistance, it would be convenient if WSR was dispersed according to an exponential function. If WSR dispersal was realistically approximated by an exponential type kernel, understanding and projecting WSR impacts would be relatively tractable, as the constant rate of disease spread after an initial and short period of increasing velocity [15] could be approximated by a diffusion rate [49]. Diffusion rate projections are often applied in invasive species models [7,50,51] as they are in plant epidemiological models [48,49,52]. Furthermore, the exponential dispersal kernel simulations consistently projected a dominant influence of the at-risk population disease resistance properties (susceptibility) on the end epidemic severity (Figure 2). Conceptually, disease resistance properties and host density within the at-risk population are the fundamental underpinnings and assumptions of effective modern disease management tactics, such as quarantine zones, vaccinations, and ring culls [1,[53][54][55]. Although, the prioritization of the at-risk population to control disease outbreaks is intuitive, as on-the-ground approaches often prioritize protecting and modifying the at-risk population to contain and dampen the impacts of any disease outbreak, it may be only a partial solution [55]. In contrast to the relatively consistent and potentially straightforward projections of the exponential dispersal kernel simulations, the modified power law dispersal kernel projections were markedly variable and not intuitively predictable over the range of conditions evaluated. Modified power law kernel simulations suggested that the outbreak may generate a strong and dominant influence on the resulting epidemic severity when compared to the at-risk population, especially at low outbreak disease levels. These disease projection results are counter to most disease mitigation approaches which are directed towards treating and prioritizing the at-risk population (e.g., quarantines, vaccinations, ring culling). There is theoretical [56], empirical [10,45,47] and in silico support [45] for a dominant influence of the outbreak on the epidemic severity in the at-risk population, but the mechanisms governing this phenomenon are not yet well-understood. However, at higher outbreak disease levels, >1% of the total possible infections at the outbreak onset, the power law dispersal kernel projections suggested that the outbreak and at-risk population susceptibility properties may exert a roughly equal contribution to the end epidemic severity and at higher outbreak disease levels the at-risk population was projected to have a greater influence than the outbreak (Figure 2A vs. Figure 2B,C). These results suggest that the contributions of the outbreak and the at-risk population may be highly context dependent and challenging to predict if non-exponentially bound, heavy-tailed dispersal kernels are used to more realistically account for the potential of long-distance dispersal. This suggests that with increasing biological accuracy in the dispersal kernel, epidemic projections will likely become both more complex, context dependent, and unfortunately maybe necessarily nuanced. The landscape scenarios I used in this study were straightforward ( Figure 1) and relatively simple compared with the host spatial complexity of cultivar mixtures, intercropping, and a patchwork landscape mosaic of variably sized agricultural fields featuring different wheat cultivars interspersed with fields/habitats without WSR host plants. Regardless, Life 2022, 12, 1727 9 of 11 when applying a modified power law dispersal kernel, which empirical data strongly supports over an exponential dispersal kernel [17,18,32], it is clear that the most biologically accurate of the two dispersal kernels yields unintuitive WSR disease projections even though there was no anisotropic disease dispersal (a well-known feature of wind dispersed pathogens [49,57]), homogenous host distribution, and host plants with invariant physiological states (both of which are not biologically true). As attractive, convenient, and readily interpretable as the disease projections from the exponential model appear to be, such biologically inaccurate models have the distinct potential to lead epidemiological understanding and the resulting control management practices down a deceptive path.	2022-11-01T15:09:29.927Z	2022-10-28T00:00:00.000	{ "year": 2022, "sha1": "ca8cc6e953fc7f3df68991d3c829148a9f01e4cb", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2075-1729/12/11/1727/pdf?version=1666948475", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "ffb3f3ab40ed214319f92cf827509b1a8360fd99", "s2fieldsofstudy": [ "Environmental Science" ], "extfieldsofstudy": [] }
30221968	pes2o/s2orc	v3-fos-license	SensorHUB: An IoT Driver Framework for Supporting Sensor Networks and Data Analysis The Internet of Things (IoT) is transforming the surrounding everyday physical objects into an ecosystem of information that enriches our everyday life. The IoT represents the convergence of advances in miniaturization, wireless connectivity, and increased data storage and is driven by various sensors. Sensors detect and measure changes in position, temperature, light, and many others; furthermore, they are necessary to turn billions of objects into data-generating “things” that can report on their status and often interact with their environment. Application and service development methods and frameworks are required to support the realization of solutions covering data collection, transmission, data processing, analysis, reporting, and advanced querying. This paper introduces the SensorHUB framework that utilizes the state-of-the-art open source technologies and provides a unified tool chain for IoT related application and service development. SensorHUB is both a method and an environment to support IoT related application and service development; furthermore, it supports the data monetization approach, that is, provides a method to define data views on top of different data sources and analyzed data. The framework is available in a Platform as a Service (PaaS) model and has been applied for the vehicle, health, production lines, and smart city domains. Introduction The goal of the Internet of Things (IoT) is to increase the connectedness of people and things. The IoT is the network of physical things equipped with electronics, software, sensors, and connectivity that provides greater value and better service by exchanging data with the manufacturer, operator, and/or other connected devices. Each element of the network, that is, each thing, is uniquely identifiable through its embedded computing system and is able to interoperate within the existing Internet infrastructure [1]. Things in the IoT can refer to a wide variety of devices such as biochips on farm animals, heart monitoring implants, production line sensors in factories, vehicles with built-in sensors, or field operation devices that assist firefighters [2]. These devices collect useful data with the help of various existing technologies, then autonomously flow the data between other devices, and usually upload them into a data center environment for further processing. The IoT together with the collected and analyzed data can help consumers achieve goals by greatly improving their decision-making capacity via the augmented intelligence of the IoT. For businesses, the Internet of Business Things helps companies achieve enhanced process optimization and efficiency by collecting and reporting on data collected from the business environment. More and more businesses are adding sensors to people, places, processes, and products to gather and analyze information in order to make better decisions and increase transparency [3]. Various sensors drive the IoT ecosystems and make the things active elements. They are our eyes and ears to what is going on in the world. IoT sensors are also expected to generate large amounts of data from diverse locations and various domains; for example, weather, transportation, and communication data can be aggregated, analyzed, and utilized for support different fields, for example, smart city services [4]. Undoubtedly, the Internet of Things has reached and is about to dominate several domains. Top industries investing in sensors and utilizing data collected by them are as follows (some of them are still in active research phase, because of 2 International Journal of Distributed Sensor Networks technical challenges and economical issues, but others are already being implemented) [5]. (i) Energy and Mining. Sensors continuously monitor and detect dangerous carbon monoxide levels in mines to improve workplace safety. (ii) Power & Utilities. In the past, and mostly today, power usage is still measured on a yearly basis. However, Internet-connected smart meters can measure power usage every 15 minutes and provide feedback to the power consumer, often automatically adjusting the system's parameters. (iii) Transportation and Vehicles. Sensors planted on the roads, working together with vehicle-based sensors, are about to be used for hands-free driving, traffic pattern optimization, and accident avoidance. (iv) Industrial Internet (Industry 4.0). A manufacturing plant distributes plant monitoring and optimization tasks across several remote, interconnected control points. Specialists once needed to maintain, service, and optimize distributed plant operations are no longer required to be physically present at the plant location, providing economies of scale. This is one of the areas where significant improvements are expected in the near future. (v) Hospitality and Healthcare. Electronic doorbells silently scan rooms with infrared sensors to detect body heat, so the staff can clean when guests have left the room. Electrocardiography (ECG) sensors work together with patients' smartphones to monitor and transmit patients' physical environment and vital signs to a central cloud based system. (vi) Retail. Product and shelf sensors collect data throughout the entire supply chain. They often provide from dock to shelf logs. Predictive analytics applications process these data and optimize the supply chain [6]. (vii) Technology. Hardware manufacturers continue to innovate by embedding sensors to measure performance and predict maintenance needs. (viii) Financial Services. Telematics allows devices installed in the car to transmit data to drivers and insurers. Applications like stolen vehicle recovery, automatic crash notification, and vehicle data recording can minimize both direct and indirect costs while providing effective risk management. Each of these sensor types provides significant benefits for the targeted domains; furthermore, based on the actual trends, we expect that they will support even more sophisticated use cases, making the urban life and smart spaces more livable, safer, and cost effective. Analysts expect that 50 to 100 billion devices will be connected to the Internet by 2020. According to a BBC Research report [7], the global market for sensors was valued at $79.5 billion in 2013 and is expected to increase to $86.3 billion in 2014, $95.3 billion in 2015, and nearly $154.4 billion by 2020, a compound annual growth rate of 10.1% over the five-year period from 2015 through 2020. The IoT is on the right way to be a major source of big data, contributing massive amounts of streamed information from billions of devices and sensors. Typical IoT applications that produce big data include vehicles and transportation, meteorology, experimental physics, astronomy, biology, and environmental science. For example, a Boeing jet generates 20 TBs of data every hour during a flight. Airliners have more than 300,000 sensors on board constantly generating data streams. Indeed, machine-to-machine (M2M) communication generates enormous amounts of Internet traffic. The availability of massive amounts of information streaming from billions of IoT devices inevitably and justly requires appropriate handling methods and techniques; furthermore, in certain cases, the sensitivity of the data brings up security and privacy concerns as well [8]. The SensorHUB framework is a collection of different technologies and is assembled as a tool chain to support IoT related development: collecting the sensor data, transmitting, processing, analyzing, and supporting the utilization for different purposes. The power and the uniqueness of the solution is that the framework is designed to be available via the Platform as a Service (PaaS) model; that is, server side development, including data management and processing, reporting, push notification, data monetization, are available via a web browser. Currently, Integrated Development Environments (IDEs) still require to install them, follow updates, and usually cover only certain parts of the whole data management process addressed by the SensorHUB. A further advantage of the SensorHUB is that it makes it possible to develop and reutilize domain specific software blocks, for example, components for the health or the vehicle domain that are developed once and built into different applications. The framework makes these available by default and provides various features to support developers working in this field. In summary, SensorHUB provides both a method and an environment to support IoT related application and service development and, furthermore, emphasizes the importance of a data monetization approach, that is, providing the methods to define data views on top of different data sources and analyzed data. In this way, comparing to the available methods, environments, and frameworks, SensorHUB is a novel approach for more effective development in an emerging area. The rest of the paper is organized as follows. Section 2 provides the background and the related work and highlights the unique capabilities of the SensorHUB framework. Section 3 introduces our SensorHUB framework in depth. We discuss the architecture, components, and capabilities of the framework. Section 4 discusses the SensorHUB implementation, the VehicleICT platform. Section 5 introduces further projects utilizing the SensorHUB framework: the URBMOBI (Urban Mobile Instruments for Environmental Monitoring, i.e., a Mobile Measurement Device for Urban Environmental Monitoring) project integrates a mobile measurement unit for operation on vehicles in urban areas with data postprocessing, inclusion in enhanced environmental models and visualization techniques for climate related services, environmental monitoring, planning, and research needs. International Journal of Distributed Sensor Networks 3 The SOLSUN (Sustainable Outdoor Lighting & Sensory Urban Networks) project demonstrates how intelligent city infrastructure can be created in a cost-effective and sustainable way by reusing existing street lighting as the communications backbone. SOLSUN project develops an integrated technology platform where several components of the Sen-sorHUB framework and the knowledge of the SensorHUB team are utilized. Finally, conclusions are elaborated. Background and Related Work A pillar of the Future Internet, the Internet of Things, will comprise many billions of Internet-connected objects or "things" that can sense, communicate, compute, and potentially actuate, as well as have intelligence, multimodal interfaces, physical/virtual identities, and attributes. The IoT is an enabler and often driver to many application domains including production lines, supply chain management, transportation and logistics, aerospace, and automotive. A world is saturated with "things" that form diverse and heterogeneous networks with overlapping capabilities in massively distributed IoT based systems; therefore it is important to efficiently utilize resources, including power efficiency and sensor data based capabilities. Usually, an IoT has a radio that can actively participate in wireless signals. Wireless devices require either low power usage or frequent recharge, but the ease of their installation, the possibility of their free movement, and the available appropriate technologies made them popular. IoT wireless protocols are designed to accomplish some basic services such as operating on low power, using low bandwidth, and working on a mesh network. Devices in a mesh network connect directly with one another and pass signals like runners in a relay race. It is the opposite of a centralized network. Some devices work on the 2.4 GHz band, which is also used by Wi-Fi and Bluetooth. Today, no wireless technology has a dominant market share in IoT applications. Based on Gartner reports [9], more than 10 IoT wireless technologies will "get significant traction" in IoT applications. These wireless technologies include cellular, satellites, and further communication methods. The reason is that no wireless technology will meet every need in every circumstance, simply because they will be too diverse and often contradicting. A connected car, for instance, will use a cellular network to contact our home network [10]. A popular IoT device armed with several sensors is the connected car. Typically, a connected car made after 2010 has a head unit, in-car entertainment unit, and in-dash system with a screen from which the operations of the connections can be seen or managed. Available functions include audio playing, navigation, roadside assistance, smartphone apps, voice commands, parking apps, engine controls, and fault diagnosis [11]. The state of automotive developer programs in 2014 starts with the following phrase: "Car apps: the next big thing." Greg Ross, Global Director of Infotainment Strategy and Alliances at General Motors, explained the following: "Creating an app environment was a way to let our cars be more personalized, stay more current, have the latest content and stay fresher longer, and frankly it must be the same way that your smartphone gets better and fresher and more personalized over time" [12]. The study highlights that there are four different ways to develop apps for cars. (i) We can develop in-vehicle infotainment apps either on the head unit (apps running on the in-car information system, that is, dashboard) or (ii) via a smartphone link (the car as a smartphone accessory, where apps run on a mobile device). (iii) We can access vehicle data and interact with the car via a remote API (remotely control the car and access vehicle data) or (iv) using a Bluetooth OBD-II dongle (use the OBD port to access vehicle data) [13,14]. The OBD-II port has been mandatory in vehicles in industrialized markets for more than a decade. Currently, there are more than 150 apps using OBD-II port in the Google Play store. With this approach, the application does not run "in the car" at all, but on a smartphone, in the cloud, on a computer, or on another device. OBD-II ports generally do not provide a possibility to control the car. The platform introduced in this paper also follows the OBD-II based data collection model. According to the International Organization of Motor Vehicle Manufacturers (OICA), 84 million new vehicles were produced in 2012. This number includes consumer cars and professional vehicles like trucks as well. However, only a minority of these are "app-enabled" models. While there are more than a billion vehicles in use worldwide, making all these cars connected would require equipping them with new capabilities. The introduction of Apple's CarPlay [15], Google's Open Automotive Alliance [16], and Microsoft's Windows in the Car [17] could be the first reasonable step on the field. These three players have a deep expertise in fostering vibrant ecosystems, building developer communities, and enabling developers to experiment and discover new use cases. Of course, there is now a realistic possibility that these platforms will beat the existing car app platforms, just as they did in the smartphone world, but they are about to provide an alternative way for the application developer community [12]. The connected car area is currently an actively researched field. There are several approaches and solutions, for example, [18][19][20]. Vehicle cloud computing and vehicular networking are also emerging research fields, which focus on the usage of cloud computing for vehicular networks. Carmakers, that offer platforms for application development, typically do not make their solutions available on older models. Some car makers announced that they are working on changing this habit. According to Gartner's predictions, by 2020 in mature automobile markets more than 80% of newly sold vehicles will be equipped with connectedvehicle functionality. IHS Automotive forecasts that, by 2020, there will be 152 million connected cars. Neither prediction specifies how many of those connected cars will be open for 3rd party apps. This is the point where developments, such as the VehicleICT platform (an implementation of the SensorHUB platform targeting the vehicle domain), arise. Such solutions make car application development available 4 International Journal of Distributed Sensor Networks for both new and older car models and offer this for a wide range of developers. The implementation of the VehicleICT platform has facilitated achieving a clearer SensorHUB architecture. The VehicleICT platform utilizes the capabilities of the Sen-sorHUB; therefore, the VehicleICT platform components test the various functions of the SensorHUB framework. The main advantages of the SensorHUB framework are the PaaS model (i.e., the development features are available in a web browser as a service) and the capability for developing and utilizing domain specific software components. The strength of the framework is that it covers the whole data collection, analyzing and reporting process; furthermore, it supports the data querying, that is, data utilization in different ways, to make the data available and utilizable for various third-party purposes. The framework utilizes the most powerful open source data management, data processing, reporting, and further technologies and provides a unified tool chain for IoT related application and service development. In this way it provides a novel approach compared to the available frameworks, development environments, and methods. We utilized the following technologies and components during the implementation of the SensorHUB: (i) Node.js [21] is applied as a cross-platform runtime environment for server side applications. It provides an event-driven architecture and a nonblocking I/O API that optimizes an application's throughput and scalability. (ii) Apache Hadoop [22] is used as a software framework for distributed storage and distributed processing of large data sets on computer clusters built from commodity hardware. It consists of a distributed file system (HDFS) and a resource management platform (YARN); furthermore, it provides a basis for a great deal of purpose-built frameworks, such as Apache Spark, Apache Hive, and Cloudera Impala. (iii) Apache Spark [23] as a high-performance cluster computing framework makes possible distributed in-memory data processing. We use Spark's highlevel functional API for data processing and Spark Streaming for effective event-processing. (iv) Apache Hive [24] is applied as a data warehouse infrastructure built on top of Hadoop. It provides an SQL interface for data stored on HDFS. We use it for ETL (extract, transform, and load) batches, which require high throughput instead of low latency. (v) Hive is inherently batch-oriented due to the underlying MapReduce framework [25]. We use Cloudera Impala [26] as a massively parallel processing SQL engine that bypasses the cluster manager component of Hadoop and deploys its own processes on the cluster nodes to manage data. Together with the Parquet columnar storage format, it enables fast, interactive analytic queries against the data stored on HDFS. The SensorHUB framework is developed in an incremental and iterative way: the features and areas covered by the framework are required by one or more projects; they have been developed based on exact requirements and then abstracted to the framework level to make them reusable. This means that those parts of the framework are better worked out, which support one or more projects. Some parts of the framework, for example, different algorithms, have proof of concept implementations, but at their first application, some of them will require manual corrections. The SensorHUB Framework To realize the IoT vision of bringing technology to people anytime, anywhere, with any device, service, or application, not only must users be aware of their devices' capabilities but also the "things" must be aware of users' activities, preferences, and context. The SensorHUB concept provides a framework and tools to support application domain specific service development. The architecture of the SensorHUB concept is depicted in Figure 1. The whole system contains the following areas: (1) Sensors, data collection, local processing, client side visualization, and data transmission (bottom left). (2) Cloud based backend with big data analysis and management (bottom right). 3.1. Overview. Sensors cover different domains: health, smart city, vehicle, production line, weather, and others areas. Local processing and data transmission make up a local platform, which performs core services, that is, data collection, data aggregation, visualization, secure communication, and data transmission. This component also provides information as a local service interface for different applications. The cloud component provides the historical data storage, big data management, domain specific data analysis, extracttransform-load (ETL) mechanisms, and data query interface. Its architecture was designed specifically for cloud deployments, although it can also be deployed on conventional server instances. In the core, there is a microservice repository, which holds the implementation of the different services. The most notable domain-agnostic services are the data ingestion service and the general querying service. Among the domain specific services are the push notification service, which is applicable in all domains that have smartphones on the client side, and the proximity alert service, which can be used to determine if the user is located inside a noteworthy area and is extremely useful in the vehicular domain. The fourth layer comprises applications that implement the specific user-facing functionalities. From this point of view, it might seem that those are the components designed in the last phase, but in fact, the applications are what drive the design of the SensorHUB framework's architecture. These data driven applications, independently of their purpose, eventually face the same problems repeatedly. Without the SensorHUB framework, applications would have to find a way to collect their data (client side code on the sensors), to store them reliably in large amounts and in a scalable way (ingesting the data into a database with all the related difficulties), to transform them into a format that makes it possible to access them either for analytic purposes or present them on a dashboard (the contemporary problem of utilizing big data), and also to act on them in time, when the information is still relevant (building a stream processing pipeline). Solving these problems is not at all trivial and can account for the majority of the development effort if done one by one for every different application. The main purpose of the SensorHUB framework is to function as a platform for these applications, providing the implementation of the previously described tasks, so the application developers can focus on the domain specific problems they intend to solve. The implementation includes client side software components, which make it possible to collect sensor data with ease; a data processing pipeline that aids tasks from the data ingestion to the visualization of the data in an efficient, scalable way; and also several domain specific components, which are not as commonly needed as the data processing pipeline itself but are also reusable in different applications. Architecture. Given the scale the framework needs to operate on, handling the data of multiple data-intensive applications, we designed it to be deployed in cloud environments from the beginning. That is most apparent in the way we organized the different functionalities into microservices. Microservices are lightweight server components that focus on a single task, in contrast to monolithic server applications. This approach not only makes the services more maintainable, easier to develop independently of each other, and replaceable, but also leads to components that boot fast, which is an essential requirement when deploying to the cloud, as new instances must be fired up on the run when load increases. Most of the framework's microservices are built based on the Node.js framework, because it is lightweight, excels at IO-heavy tasks, and promotes agile development. We made the different microservices accessible through a Service Bus, which unites the microservices into a cohesive interface and hides all the service instantiation details from the applications. Another responsibility of the Service Bus is to function as a load balancer and deploy new instances of a single service in the cloud in case a specific service is overloaded. It is important for the implementation to enable the rapid launch and transparent breakdown of the microservices. Booting up a Node.js instance is relatively quick, and Service Inst. Service Inst. Service Inst. Service Inst. Service Inst. by keeping the services stateless, the load balancing task is straightforward in a cloud environment. Figure 2 provides the implementation structure of the core backend services as microservices. Service The data ingestion and the data querying microservices are the two endpoints of an important module, which is responsible for big data management. Data are ingested into a cluster of machines running Hadoop [22]. Data ingestion on the cluster side itself can become a bottleneck, so a separate load balancing mechanism is applied to the data ingestion nodes. Raw data are loaded into the Data Lake, which is a single repository for all the applications' data. Data loaded into the Data Lake are never modified, only read by the different applications. During data ingestion, the control of the schema is intentionally absent from the framework. This approach gives flexibility to the framework, as schema must be forced on the data by the applications themselves, which can enforce application-specific properties on the data. Figure 3 introduces the data ingestion process. Data Processing. Although flexibility is an asset, in most cases the schema is known at the time of data ingestion. That is why we took a hybrid approach by providing an ETL engine, which application developers can configure to load their data into one of the supported query-optimized data stores. Depending on the applications' needs, the data store can be one of the following (Figure 3): (i) A compressed, partitioned, columnar data store, implemented on a massively parallel processing engine (Cloudera Impala with Parquet files), which is efficient for analytic query patterns. (ii) A NoSQL data store (Apache HBase), which should be used if the main goal is data retrieval and especially data modification and not complex analytic queries, for example, when client applications want to display historic data to the end users. (iii) A traditional relational database (MySQL), which has the advantage that it is well-known to developers, has many good associated tools, and scales well for medium-sized services. (iv) Data can be piped into a stream processing algorithm, which can be used to detect anomalies in the data and send alert messages directly to the client applications. Alerts are sent by the push notification service of the SensorHUB framework. The data are also available in raw format in the Data Lake. As data in the Data Lake are never modified once uploaded, application developers can always access them with arbitrarily complex processing algorithms or by providing their own custom ETL. These four standard formats, supplemented by the capability of defining any custom processing, enable developers to focus on the data at the abstraction level that best fits their needs, contributing to the ease of development. Further advantage of this hybrid approach is that, for the data processed with the configurable ETL engine, we were able to create a standard query interface, because in this case, schema is known by the framework. Applications, which do not utilize the ETL engine, must handle the query interface themselves. This is a reasonable tradeoff between customizability and the ability to use general services provided by the framework. Domain Specific Applications. On top of the platform, there are the domain specific applications, web and mobile applications, and services. Special types of services are customized reports, data monitoring solutions, dashboards, and further business intelligence solutions. As the platform itself was designed to be deployed on a backend infrastructure of an internal network, it is recommended that these applications use their own web servers to utilize the platform's capabilities, although it is also possible to simply open the internal ports to client applications. We strongly advise against the latter, because this approach poses a security risk, as internal microservices are only prepared to authorize requests that are coming from a relatively safe, firewall protected environment and not from the outside world, where stronger authentication and authorization methods are required, which are the responsibility of the application-specific web servers. Figure 4 introduces a possible deployment, where the Hadoop Cluster and the SensorHUB platform are deployed on servers on an internal network, and the different user-facing services deploy their own web servers. An example would be an application that uses smartphones to collect data and provide services to the users. These smartphone applications would directly connect to their own web server, knowing nothing about the SensorHUB framework. On the other hand, the web server would wrap the services of the underlying framework and glue them together in a way best fitting for the application. The greatest strength of the SensorHUB framework is that it enables these web servers to remain a thin layer. In the absence of the framework, every single application would need to implement its own version of the infrastructure that is shown in the internal network box in Figure 4. In many of our SensorHUB utilizations, a smartphone running Android OS serves as a bridge between a sensor element of a distributed sensor network and the infrastructure behind. As many of these sensors have no direct Internet access but are capable of communicating using Bluetooth, an Android smartphone with the capability of Bluetooth connection and mobile Internet access is able to serve this purpose. Clients Side Support. In order to support the client application development, client side services are available. The client side services are implemented on the Android smartphone platform and available as an application library. This library encapsulates these services and provides them as independent building blocks. The first part of these modules is the client side counterpart of the platform services available on the infrastructure side. These are client side utilities that support these services on the client side, such as transparent push notification handling and device registration, or data querying. The second part of client side modules is the client utilities. These modules provide services for common client side domain-independent features, including reliable networking, secure communication, and easier integration with social services ( Figure 5). 3.6. Summary. SensorHUB is a general concept with a core platform implementation. We provide different realizations (domain specific software components), that is, utilizations of the SensorHUB platform. The results are different specialized platforms targeting a selected area. Such platform is the VehicleICT platform for the vehicular area. The next section discusses the VehicleICT platform and Section 5 introduces further SensorHUB based realizations. The VehicleICT Platform The VehicleICT platform is an implementation on top of the SensorHUB framework targeting the vehicle domain. The implementation of the VehicleICT platform helped to distill the architecture of the SensorHUB. The VehicleICT platform utilizes the capabilities of the SensorHUB and provides a vehicle domain related layer with several reusable components and features. This means that the VehicleICT platform itself can be considered as a test environment that verifies the different aspects of the SensorHUB framework. The idea behind the VehicleICT platform was to identify a reasonably rich set of functionalities that typical connected car applications need and then to implement and test these functionalities and finally offer them as building blocks in a centralized manner. The approach enables application developers to focus on their domain related logic. By using the building blocks, application development becomes more efficient and leads to more stable software artefacts. Applications and services in the connected car domain can be divided into three separate parts: (i) the sensors, (ii) the local processing and visualization, and (iii) the background processing and analytics. The VehicleICT platform meets developers' needs in all three of aforementioned areas [27]. Figure 6 introduces the architecture of the platform. to access the sensor data: through the OBD-II port or by connecting directly to the CAN bus. The OBD-II port can be found on the panel of every car manufactured in the last decade, but until recently, it was mainly used by repair shops to detect faults in the internals of the vehicle, even though it has much more potential in it. Nowadays, small inexpensive devices are available, which can access sensor data through the OBD-II port and do not require special expertise to install, which makes them a convenient choice for average customers. The OBD-II port is connected to the vehicle's CAN bus; however it enables access to only a limited set of vehicle sensors. That is why a special device has been developed, which can be connected directly to the CAN bus, bypassing the restrictions of the OBD-II. The downside of this solution is that it requires an expert to install the device, so use cases are restricted. Both devices broadcast the data they collect via Bluetooth. Local Processing and Visualization. A background component running on the user's device (smartphone, tablet), which we call the Platform Core, is responsible for abstracting away the differences in the previously described methods. This service has no user interface but is available for connected client applications through an API (Platform Library). This service is a singleton, as one instance is responsible for serving all the connected client applications in parallel. Although every client application has a platform implementation in the Platform Library, the Platform Wrapper is responsible for redirecting the communication to the singleton instance and wraps the interprocess communication details. As the Platform Core implementation is located in the Platform Library, no additional application or driver is needed to use the Platform Core. A client application, as a user of the framework, connects to the Platform Core and requests some data, for example, the engine RPM (the frequency of its rotation). The Platform Core connects to the available collector device via Bluetooth, acquires the same data from the vehicle sensors, and delivers it to the client application. The application does not need to concern itself with the origins of the data. Figure 7 shows the main components an Android Client application interacts with. The Platform Core is prepared to serve an arbitrary number of applications at the same time. The standard operation involves providing a set of data to applications periodically, so it is possible to optimize the access to International Journal of Distributed Sensor Networks the sensor devices by requesting data only once, even if multiple applications need them. The communication with the infrastructure side is implemented by the client application. However, SensorHUB client utilities are available for client application developers in order to support this process. These consist of networking utilities providing access to the infrastructure via REST (REpresentational State Transfer) and push notification utilities for infrastructure initiated communication. They let developers focus on the API design instead of the implementation of the communication. To promote the uniformity of applications based on the framework, we also created a UI library with domain specific UI components, such as drag-n-drop speedometers. Figure 8 depicts screenshots from the proof of concept mobile application of the VehicleICT platform. The first screen is a dashboard for further navigation. On the second screen, the Board Computer can be found, which contains the most frequently used indicators like vehicle speed, engine RPM, and ambient temperature. This screen displays the current values on an interface designed to be accessible during driving. The third screen contains the Engine Details view, which presents all available vehicle information (except those on the previous screen), in a simple form, more suitable for diagnostics. The last screen is the Eco Driving view, which displays fuel consumption and CO 2 emission related information, utilizing the services provided by the infrastructure. Background Processing and Analytics. On the infrastructure side, the data collected from the vehicles and forwarded by the client applications are aggregated, processed, and loaded into the SensorHUB based data store. The data are utilized and monetized based on the capabilities discussed in Section 3. SensorHUB based Projects The SensorHUB concept continuously evolves due to its usage in both R&D and industrial projects. The concept has been defined after the first few similar IoT projects driven by the close requirements and solutions of the different projects. This is a natural process in the software industry that, having solved the same or similar task more than twice, we are about to work out a solution that can be utilized in different projects only by configuring the general components. The incremental and iterative development of the framework is driven by both the introduction of the new IoT domains and also the evolving end user and corporate requirements targeting the data processing methods, reporting, and data monetization ways. VehicleICT was the first project, where both the client and server parts of the SensorHUB framework have been utilized. In certain cases, based on the actual conditions and requirements, we apply only a part of the whole framework, that is, either the data collector sensor area (client side) or the backend component with extensive reporting. Within the frame of two EIT (European Institute of Innovation & Technology) Climate-KIC [28] projects we utilize primarily the sensors related client part and the data upload components of the framework. These Climate-KIC projects are referred to as URBMOBI and SOLSUN. In addition, in the SOLSUN project we are developing several web based and reporting components on top of the SensorHUB platform. The experience shows that the framework increases both the ease of development of the certain features and also the quality of the resulted software components. Furthermore, the extension of the services and the maintenance of the source code and the components of the solutions are also supported by the framework methods. In summary, the framework with integrated solutions (data management, analysis, reporting, and push notification) moves the development activities to a higher abstraction level and provides an up-to-date professional environment for the developer teams. The URBMOBI (Urban Mobile Instruments for Environmental Monitoring, i.e., a Mobile Measurement Device for Urban Environmental Monitoring) project integrates a mobile measurement unit for operation on vehicles in urban areas (i.e., local buses and trams), with data postprocessing, inclusion in enhanced environmental models and visualization techniques for climate related services, environmental monitoring, planning, and research needs. URBMOBI is a mobile environmental sensor that (i) provides temporally and spatially distributed environmental data, (ii) fulfills the need for monitoring at various places without the costs for a large number of fixed measurement stations, (iii) integrates small and precise sensors in a system that can be operated on buses, trams, or other vehicles, (iv) focusses on urban heat and thermal comfort, and (v) aims at providing climate services and integration with real-time climate models. The URBMOBI solution provides a novel product that integrates state-of-the-art sensors for environmental variables embedded in a system that allows mobile usage and data handling based on geolocation technology and data transmission by telecommunication networks. Sensors can be operated on buses, trams, taxis, or similar vehicles in urban areas. The data are geocoded and postprocessed depending on the type of variable, location, and application. Furthermore, the data are integrated into real-time models on climate and/or air quality relevant quantities providing climate services and environmental data for a wide range of applications. URBMOBI is utilizing the SensorHUB framework in data collection, local processing (data aggregation), and data transmission. On the server side, URBMOBI measurements are combined with atmospheric models in order to improve spatial coverage and calculate additional parameters (thermal comfort). The data are analyzed with a climate domain related powerful tool. A part of the SensorHUB architecture has been redesigned and improved based on the experience collected at the URBMOBI project. As a result we have obtained a clearer framework architecture. URBMOBI project has been worked out between 2013 and 2015 by the following consortium: RWTH Aachen University (Germany), Netherlands Organisation for Applied Scientific Research TNO (Netherlands), ARIA Technologies (France), Budapest University of Technology and Economics (Hungary), and Meteorological and Environmental Earth Observation (Italy). The SOLSUN (Sustainable Outdoor Lighting & Sensory Urban Networks) project is about to demonstrate how intelligent city infrastructure can be created in a cost-effective and sustainable way by reusing existing street lighting as the communications backbone. We apply different technologies and methods to reduce energy consumption at the same time as turning streetlights into nodes on a scalable network that is also expandable for other applications. Sensors capture data on air pollution, noise pollution, and traffic density; information gathered is used to address traffic congestion, another key contributor of greenhouse gas emissions in cities. SOLSUN project develops an integrated technology platform where several components of the SensorHUB framework and the knowledge of the SensorHUB team are utilized. The project brings together a strong core of public, private, and academic partners with the combined expertise to develop outcomes that can be exploited on a global scale. The project is carried out between 2015 and 2017 by the following partners: Select Innovations Limited (UK), British Telecommunications Plc (UK), Municipality of the City of Budapest (Hungary), PANNON Pro Innovation Services Ltd. (Hungary), and Budapest University of Technology and Economics (Hungary). The technology will initially be tested at British Telecommunications' R&D headquarters on a lighting installation, and later a demonstration will be delivered to the streets of Budapest. Sensor and sensor network development is supported by the SensorHUB framework; the data analysis is mainly carried out on British Telecommunications' Data HUB architecture. According to the predictions, up to 100 billion devices will be connected to the Internet by 2020. The SOLSUN technology is designed to be scalable to cope with the growing demand for networked devices. The system can cater for 254 device types with 65,000 devices in one category; multiple protocols are embraced with data sent back to a scalable cloud based Cluster Controller, with no upper limit on the amount of Cluster Controllers. This enables providers to carry on using their preferred protocol but still benefit from a web based front end and/or application connection. To ensure scalability, connections are made through stand-alone adapters; multiple adapters can be distributed and software can run on many servers with no single point of redundancy. Besides the connected car domain (VehicleICT) and the smart city domain (Climate-KIC projects), we are currently addressing two more domains, namely, the health and the production line (industrial Internet). The architecture is similar; that is, data are collected with domain related sensors, locally processed and utilized, furthermore uploaded, and analyzed, and services are driven by the distilled data. These projects develop domain specific solutions on top of the SensorHUB. Our experience shows that these IoT projects, based on the utilized components and both the way and results of the development, validate the SensorHUB approach. Conclusions "Data is the new oil." We often meet with similar statements. IoT based data collection, data transmission, big data management, trusted cloud, and privacy issues are the main challenges of this area. Frameworks helping the companies, research groups, and students contribute to this ecosystem and the future design and development platforms. Based on the realized developments and ongoing project activities we can state that SensorHUB is such a framework, especially in the following ways: (ii) It provides a basis and effective conditions not only for application development, but also for core R&D activities, for example, evaluating and comparing network traffic, data analysis, data security algorithms, and software development solutions in a real environment. Beside the industrial purposes, the framework is utilized by MSc students in project laboratories and by PhD students for proof of concept developments, for example, to prepare their measurement environments. (iii) The framework is a common basis for data driven development: both the client and the server side support effective application and service development. The concept can be configured for different domains (application areas). The sensors and data collection requires domain specific development. However, the client and server components of the framework provide methods for data collection, local processing and data visualization, data transmission to the server side storage, data analysis, and using the information in different ways to build services and applications. Currently addressed domains are healthcare, manufacturing and production lines, smart city applications, vehicular area, and industrial Internet (Industry 4.0). SensorHUB is a unified tool chain for IoT related application and service development. Furthermore, it emphasizes and supports the data monetization; that is, it provides the method to define data views on top of different data sources and analyzed data. The framework is available in a Platform as a Service (PaaS) model. The paper has discussed the motivation, the objectives, and also the application areas and domains of the Sen-sorHUB framework. Based on the current industrial trends, requirements, and needs, SensorHUB framework is a data monetization enabler. The framework supports the collection of the various sensor data, enables the processing and analysis of the data, and makes it possible to define different views on top of the data combined and compiled from different data sources. These data views and collections of datasets are referred to as monetized data for various purposes, for example, supporting decision making and running smart city services.	2018-04-03T02:57:17.042Z	2015-07-01T00:00:00.000	{ "year": 2015, "sha1": "e6e0b194e9df42292e3bd8dc44f1e2de338e7586", "oa_license": "CCBY", "oa_url": "https://doi.org/10.1155/2015/454379", "oa_status": "GOLD", "pdf_src": "Sage", "pdf_hash": "4e84ebc3aec9195a03223002da9602e82353f70d", "s2fieldsofstudy": [ "Computer Science", "Engineering", "Environmental Science" ], "extfieldsofstudy": [ "Computer Science" ] }
33597451	pes2o/s2orc	v3-fos-license	Point shear wave elastography method for assessing liver stiffness AIM: To estimate the validity of the point shear-wave elastography method by evaluating its reproducibility and accuracy for assessing liver stiffness. METHODS: This was a single-center, cross-sectional study. Consecutive patients with chronic viral hepatitis scheduled for liver biopsy (LB) (Group 1) and healthy volunteers (Group 2) were studied. In each subject 10 consecutive point shear-wave elastography (PSWE) measurements were performed using the iU22 ultrasound system (Philips Medical Systems, Bothell, WA, United States). Patients in Group 1 underwent PSWE, transient elastography (TE) using FibroScan (Echosens, Paris, France) and ultrasound-assisted LB. For the assessment of PSWE reproducibility two expert raters (rater 1 and rater 2) independently performed the examinations. The performance of PSWE was compared to that of TE using LB as a reference standard. Fibrosis was staged according to the METAVIR scoring system. Receiver operating characteristic curve analyses were performed to calculate the area under the receiver operating characteristic curve (AUC) for F ≥ 2, F ≥ 3 and F = 4. The intraobserver and interobserver reproducibility of PSWE were assessed by calculating Lin’s concordance correlation coefficient. RESULTS: To assess the performance of PSWE, 134 consecutive patients in Group 1 were studied. The median values of PSWE and TE (in kilopascals) were 4.7 (IQR = 3.8-5.4) and 5.5 (IQR = 4.7-6.5), respectively, in patients at the F0-F1 stage and 3.5 (IQR = 3.2-4.0) and 4.4 (IQR = 3.5-4.9), respectively, in the healthy volunteers in Group 2 (P < 10). In the univariate analysis, the PSWE and TE values showed a high correlation with the fibrosis stage; low correlations with the degree of necroinflammation, aspartate aminotransferase and gamma-glutamyl transferase (GGT); and a moderate negative correlation with the platelet count. A multiple regression analysis confirmed the correlations of both PSWE and TE with fibrosis stage and GGT but not with any other variables. The following AUC values were found: 0.80 (0.71-0.87) for PSWE and 0.82 (0.73-0.89) for TE (P = 0.42); 0.88 (0.80-0.94) for PSWE and 0.95 (0.88-0.98) for TE (P = 0.06); and 0.95 (0.89-0.99) for PSWE and 0.92 (0.85-0.97) for TE (P = 0.30) for F ≥ 2, F ≥ 3 and F = 4, respectively. To assess PSWE reproducibility, 116 subjects were studied, including 47 consecutive patients scheduled for LB (Group 1) and 69 OBSERVATIONAL S UDY consecutive healthy volunteers (Group 2). The intraobserver agreement ranged from 0.83 (95%CI: 0.79-0.88) to 0.96 (95%CI: 0.95-0.97) for rater 1 and from 0.84 (95%CI: 0.79-0.88) to 0.96 (95%CI: 0.95-0.97) for rater 2. The interobserver agreement yielded values from 0.83 (95%CI: 0.78-0.88) to 0.93 (95%CI: 0.91-0.95). CONCLUSION: PSWE is a reproducible method for assessing liver stiffness, and it compares with TE. Compared with patients with nonsignificant fibrosis, healthy volunteers showed significantly lower values. © 2014 Baishideng Publishing Group Co., Limited. All rights INTRODUCTION The prognosis and management of chronic viral hepatitis depend on the extent and progression of liver fibrosis, which constitute the most important predictor of disease outcome and influence the indication for antiviral treatment [1] . The recent guidelines for the management of hepatitis C infection from the European Association for the Study of the Liver allow the use of TE, instead of liver biopsy (LB), in patients with chronic hepatitis C for as-sessing liver disease severity prior to therapy at a safe level of predictability [12] .TE has been approved by the French National Health Authority for the evaluation of fibrosis in treatment-naïve patients with chronic hepatitis C and no comorbidities [13] . Shear wave elastography techniques have been implemented in conventional real-time ultrasound systems, and several studies have shown their accuracy in the assessment of liver fibrosis [14][15][16][17][18][19][20][21][22] .Compared with TE, these techniques have the advantage of B-mode image guidance; thus, they can allow the user to choose the best acoustic window for correctly performing an examination in real time. The aim of this study was to estimate the validity of a new point shear wave elastography (PSWE) technique by evaluating the reproducibility of measurements and the accuracy of this method in the assessment of liver fibrosis.The performance of PSWE was compared to that of TE using liver histology as a reference standard. Subjects and study design This was a single-center, cross-sectional study.All consecutive patients with chronic viral hepatitis who were scheduled for liver biopsy at the Infectious Diseases Department of Policlinico San Matteo were enrolled in the study (Group 1).Consecutive healthy volunteers were also enrolled (Group 2). The accuracy of PSWE in the assessment of liver fibrosis was prospectively estimated in consecutive patients in Group 1. LB was performed on the same day as the PSWE and TE measurements, as day-case procedures.Examinations were performed in the morning after an overnight fast.The patients' characteristics, epidemiological data and biochemical test results were recorded. The reproducibility of PSWE measurements was prospectively assessed in consecutive subjects in Group 1 and Group 2. Two expert raters (rater 1 and rater 2) independently performed 10 consecutive measurements in each subject.All the subjects were asked to fast for at least six hours prior to the examination.The intraobserver agreement was assessed by comparing the median values of all the measurements and by comparing several combinations of measurements or single measurements.The interobserver agreement was assessed by comparing the median value of the 10 measurements performed in the same subject by each rater and by comparing combinations of measurements or single measurements. Moreover, the results of liver stiffness measurements performed in patients with nonsignificant fibrosis (F0-F1) were compared to the values obtained in the healthy volunteers in Group 2. Three physicians, each of whom was blinded to the other's results, independently performed the measurements.The PSWE measurements were performed by G.F. and M.Z., and the TE measurements were performed by M.Z. and R.L. The study protocol was approved by the institutional Ethics Committee.The participants provided written informed consent. Liver biopsy LB was performed by three experienced physicians (C.F., G.M. and E.B.) using a 17-gauge modified Menghini needle (Hepafix; Braun, Melsungen, Germany).The same intercostal space used for the TE and PSWE measurements was chosen for LB.The specimens were assessed on site by a single expert liver pathologist (B.D.B.) who was blind to both the TE and PSWE results.Liver fibrosis and necroinflammatory activity were evaluated semiquantitatively according to the METAVIR system [23] .Steatosis was graded according to the method of Kleiner et al [24] as S0, steatosis in fewer than 5% of hepatocytes; S1, 5%-33%; S2, 34%-66%; and S3, more than 66%. Transient elastography TE measurements were performed using the M probe of the FibroScan ® device by two physicians (M.Z. and R.L.) with experience performing at least 50 TE procedures.During the acquisition, the patients lay in the dorsal decubitus position with the right arm in maximum abduction.The results were expressed in kilopascals (kPa).Only examinations with 10 valid measurements and an interquartile range/mean (IQR/M) < 30% for values greater than 7.1 kPa were considered reliable [2,4,25] . Point shear wave elastography The examinations were performed using the iU22 ultrasound system (Philips Healthcare, Bothell, WA, United States) with a convex broadband probe and the ElastPQ ® technique.As with other shear wave elastography methods, this technique generates shear waves inside the liver using radiation force from a focused ultrasound beam.The ultrasound machine monitors the shear wave propagation using a Doppler-like ultrasound technique and measures the velocity of the shear wave.The shear wave velocity is displayed in meters per second (m/s) or in kPa through Young's modulus E = 3 (vS 2 .ρ),where E is Young's modulus, vS is the shear wave velocity and ρ is the density of the tissue.If the amount of non-shear wave motion exceeds a threshold, the system does not display a calculation. The two raters performing the PSWE measurements (G.F. and M.Z.) had seven years and two years, respectively, of experience in real-time elastography studies.They received training in PSWE measurements for two days before the study began.The examinations were performed in the right lobe of the liver through intercostal spaces, with the subject lying supine with the right arm in maximal abduction.Using a real-time B-mode image, the rater selected a vessel-free area, at least 1.5 cm below Glisson's capsule, where a fixed region of interest of 0.5 cm × 1.5 cm was placed by moving a trackball.The pa-tients were instructed to hold their breath while the rater pressed a button that launched the data acquisition.Each rater performed 10 valid measurements, which were expressed in kPa.Measurements < 1 kPa were rejected by the raters. Sample size considerations for the accuracy of PSWE: A total sample size of 130 subjects, which included 65 subjects with the disease, i.e., a prevalence of approximately 50%, was estimated to achieve 88% power to detect changes in sensitivity and in specificity from 0.75 to 0.90 using a two-sided binomial test.The target significance level was 0.05. Sample size considerations for reproducibility of PSWE: A sample size of 100 subjects, with two observations per subject, was estimated to achieve 97% power to detect a concordance correlation of 0.95 under the alternative hypothesis when the concordance correlation under the null hypothesis was 0.90 using an F-test with a significance level of 0.05. Descriptive statistics were produced for the demographic, clinical and laboratory characteristics of this study sample of patients.The Shapiro-Wilk test was used to test the normal distribution of quantitative variables.For quantitative variables that were normally distributed, the results were expressed as mean ± SD; otherwise, medians and interquartile ranges (IQR; 25 th -75 th percentile) were reported.Qualitative variables were summarized as counts and percentages.A one-way ANOVA or the Kruskal-Wallis analysis of variance by ranks, with a Bonferroni correction, was used to analyze differences among patients undergoing liver biopsy.Pearson's or Spearman's rank coefficient was used to identify correlations between two study variables. Linear regression was used for the multivariate model.A frequency distribution was obtained to choose optimal cut-off values of PSWE and to maximize the sum of the sensitivity and specificity for different fibrosis thresholds: F0-F1 vs F2-F4 (F ≥ 2), F0-F2 vs F3-F4 (F ≥ 3) and F0-F3 vs F4 (F = 4).For TE, we used cut-off values determined in a previous study [26] .The diagnostic performance of PSWE, TE and their combinations was assessed using receiver operating characteristic (ROC) curves and an area under the ROC (AUC) curve analysis.Comparisons of AUCs were performed using the method described by DeLong et al [27] for correlated data.The Obuchowski measure was used to take into account all the pairwise comparisons between stages to minimize the spectrum effect and the risk of multiple testing [28] . Interobserver reproducibility was assessed by calculating Lin's concordance correlation coefficient (CCC) [29] .The CCC combines measures of both precision and accuracy to determine the degree of deviation of the observed data from the line of perfect concordance (i.e., the line at 45 degrees on a square scatterplot).The CCC increases in value as a function of the proximity of the were found. After corrections for gender and age, both the PSWE and TE values differed significantly between the patients with chronic hepatitis C at the F0-F1 stage (n=50) and the healthy volunteers (n = 69) (Figure 1). Liver stiffness assessment: Comparison of PSWE and TE: The median values, interquartile ranges, ranges, numbers of outliers and P values of the measurements obtained for each fibrosis stage using PSWE and TE are shown in Figure 2. data's reduced major axis to the line of perfect concordance (the accuracy of the data) and as a function of the tightness of the data about its reduced major axis (the precision of the data).CCC values range from 0 to +1.As CCC values approach 1, the measurement differences between the different raters become negligible and more consistent.The interobserver agreement was classified as poor (CCC = 0.00-0.20),fair to good (CCC = 0.40-0.75)or excellent (CCC > 0.75) [30] .The CCCs were reported with 95% confidence intervals (CIs). The data analysis was performed with the STATA statistical package (release 11. RESULTS From August 2011 through April 2013, one hundred and thirty-four consecutive subjects in Group 1 and sixtynine subjects in Group 2 were prospectively studied.Data from some patients in Group 1 have been reported in previous studies [18,26] . Performance of PSWE One hundred forty patients were eligible during the recruitment period.Six patients were excluded because they were undergoing antiviral therapy.Due to patient recruitment from our referring physicians, there were no patients with overt cirrhosis or ascites in this series of patients.One hundred thirty-four patients met the inclusion criteria.LB was performed in all the patients on the same day as the PSWE and TE measurements, and no complications were observed.The specimen length was adequate for liver histology in all but one patient.TE was feasible in all but one patient, whereas the PSWE measurements failed in five patients.The PSWE measurement failures were due to narrow intercostal spaces in four cases and obesity in one. The characteristics of the 134 patients are summarized in Table 1.The mean length of the LB specimens was 2.5 (0.78) cm.The results of the statistical analysis performed on the data from the 102 patients with chronic hepatitis C are provided hereafter. No correlations with other variables, including steatosis, The optimal cut-off values for different levels of fibrosis were determined by analyzing the ROCs for PSWE.For TE, we used cut-off values obtained in a previous study [26] .The cut-off values of PSWE and TE for each METAVIR stage, along with the AUCs, sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio and negative likelihood ratio, are presented in Table 2. Figure 3 shows the ROC curves for significant (F ≥ 2) and severe fibrosis (F ≥ 3), as well as for cirrhosis (F = 4).For staging advanced fibrosis (F ≥ 3), TE had higher accuracy than PSWE, but this difference did not reach statistical significance.The Obuchowski measures were good for both PSWE [0.80 (95%CI: 0.73-0.86)]and TE [0.83 (95%CI: 0.77-0.90)]. DISCUSSION This study was undertaken to assess the validity of Ferraioli G et al .Point shear wave elastography for stiffness assessment PSWE, i.e., the repeatability of measurements and the performance of this method.The results show that PSWE is a highly reproducible method for assessing liver stiffness because it was characterized by very high levels of intraobserver and interobserver agreement, both overall and for single measurements.Moreover, the reproducibility of the method was similar in healthy subjects and in patients with chronic viral hepatitis.Ultrasound imaging techniques are subject to user dependency; nonetheless, we observed a high interobserver agreement rate that was similar to that reported for TE [9] .Nevertheless, good interobserver agreement rates have been reported for other shear wave elastography ultrasound-based techniques, suggesting that the method itself has low variability and requires only a short period of training to be performed reliably [31][32][33] .Indeed, the benefits of image guidance will likely reduce the learning curve and the variations between measurements [33] . The results of this study show that TE and PSWE results are directly and linearly correlated with the stages of fibrosis determined using histology.Furthermore, the performance of PSWE compares with that of TE, the first available technique and the most widely accepted method for noninvasive assessment of liver fibrosis.In our series, liver stiffness did not correlate with liver steatosis in either the univariate or multivariate analysis; thus, steatosis was not a confounding variable.This result is similar to those observed in other studies using shear wave elastography techniques integrated into ultrasound systems, and this result appears to indicate that the value obtained was a true estimate of the stiffness of the liver [14,16,18,19,34] .The influence of necroinflammation on liver stiffness is controversial; some studies have found an influence [9,11,14,19,34] , and others have not [2][3][4]16,18,22] . In our series, we found no correlation between liver stiffness and necroinflammation.A positive correlation with GGT was found, which is in agreement with the results of a study by Forns et al [35] , which identified this variable as an independent predictor of liver fibrosis. The diagnostic accuracy of PSWE was similar to that reported by some other studies, which used a dif- ferent point shear wave elastography method that also included acoustic radiation force impulse (ARFI); those studies reported no improvement in accuracy relative to TE for staging liver fibrosis [14,20,21,36] .Rizzo et al [16] found that ARFI was more accurate than TE for the staging of both significant and severe liver fibrosis.However, those results were not confirmed by a recent meta-analysis that compared ARFI with TE and found comparable diagnostic accuracies of both methods for the diagnosis of severe fibrosis and a slightly but significantly higher diagnostic accuracy of TE for the diagnosis of significant fibrosis and cirrhosis [17] . PSWE is a recently developed method that is part of the second generation of ultrasound elastography methods.These methods differ from the first-generation TE in several aspects, including the generation of shear waves within the organ by a focused ultrasound beam and the capability of focusing the beam at different locations within the organ under ultrasound image guidance.These properties should improve the feasibility of stiffness measurements in obese patients and patients with ascites; they may also improve the accuracy of PSWE relative to TE.However, the current study demonstrated that neither the feasibility nor the accuracy of PSWE was higher than that of TE.This finding could be attributable to the fact that the patients in our series had a body mass index within the normal range and the absence of patients with ascites.On the other hand, compared with TE, routine ultrasound systems with an elastography technique are advantageous in that they also allow the evaluation of other parameters that are complementary to stiffness, they are highly accurate for the diagnosis of cirrhosis and they could be used to screen for focal liver lesions [20,37] . The optimal cut-offs identified for each fibrosis stage, which were based on the maximal sensitivity and specificity, were close to each other.However, the diagnostic accuracy, assessed with AUCs, was high, suggesting that the PSWE method is acceptable for staging liver fibrosis but needs to be refined.On the other hand, liver histology could be an imperfect gold standard because it is affected by intraobserver and interobserver variabilities in fibrosis assessment and represents only 1/50000 of the entire liver mass [38] .Moreover, a recent study showed that liver biopsy exhibited a relative lower level of performance compared with FibroTest and TE when evaluated similarly for the diagnosis of advanced fibrosis [39] .As was very recently reported for TE [25] , in our study, both TE and PSWE showed excellent negative predictive value for cirrhosis and very good positive predictive value for significant fibrosis.On the contrary, both techniques showed insufficient positive predictive value for cirrhosis and only fair negative predictive value for significant fibrosis. The values of stiffness obtained using PSWE in healthy subjects were significantly lower than those obtained in patients with nonsignificant fibrosis (F0-F1) based on liver histology.This result indicates that the PSWE technique, which is noninvasive and readily available in ultrasound systems, could be a useful adjunct tool when performing ultrasound examinations of the liver because this method may allow physicians to select patients who need to be further evaluated for chronic liver disease. Our study has limitations.First, the different stages of fibrosis, particularly advanced fibrosis and cirrhosis, were not equally represented among the patients in our series; almost half of the patients were at the F0-F1 stage, which may have affected the optimal cut-off values obtained with the ROC curves.This uneven distribution of fibrosis stages among consecutive patients reflects what is normally observed in clinical settings.On the other hand, the Obuchowski measure, which was used to minimize the spectrum bias, was good for both Ferraioli G et al .Point shear wave elastography for stiffness assessment PSWE and TE.Second, our study population had a low prevalence of obesity, which could be a technical limitation; thus, the applicability of these results is limited.Third, the analysis was performed in a relatively small number of patients; thus, these results need to be validated in larger studies. In conclusion, PSWE is a highly reproducible method for assessing liver stiffness.For staging liver fibrosis, PSWE compares favorably with TE.Healthy volunteers show significantly lower values compared with patients with nonsignificant fibrosis.Further studies in larger series of patients are needed to confirm these results. Figure 1 Figure 1 Box and whisker plots of stiffness values in patients at METAVIR stage F0-F1 and in healthy subjects.Box plots show interquartile range (box), median (line within box), range (whisker) and outliers (circle).Median values (IQR) and P values of differences are given.A: Point shear wave elastography (PSWE) values; B: Transient elastography (TE) values. Figure 2 Figure 2 Distribution of stiffness values according to METAVIR fibrosis stage.Median values, interquartile ranges, ranges, numbers of outliers and p values are given for each fibrosis stage.A: Point shear wave elastography (PSWE); B: Transient elastography (TE). Table 3 Characteristics of the subjects assessed to determine the reproducibility of point shear wave elastography measurements SD values represent means, and IQR values represent medians.a P < 0.05; b P < 0.01.PSWE: point shear wave elastography; BMI: body mass index.Group 1: Patients with chronic viral hepatitis; Group 2: Healthy volunteers. Table 4 Intraobserver and interobserver agreement of point shear wave elastography measurements performed by two raters CCC: Concordance correlation coefficient; CI: Confidence interval.	2018-04-03T06:14:49.727Z	2014-04-28T00:00:00.000	{ "year": 2014, "sha1": "1aa775f3d3c81efb806fad2b4c23c0796d3343b0", "oa_license": "CCBYNC", "oa_url": "https://doi.org/10.3748/wjg.v20.i16.4787", "oa_status": "HYBRID", "pdf_src": "ScienceParseMerged", "pdf_hash": "9ea4d56cb1af39828ea68d8a86fdbd22eeef0c06", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
238782494	pes2o/s2orc	v3-fos-license	Artificial Intelligence for Prediction of Physical and Mechanical Properties of Stabilized Soil for Affordable Housing Soil stabilization is the alteration of physicomechanical properties of soils to meet specific engineering requirements of problematic soils. Laboratory examination of soils is well recognized as appropriate for examining the engineering properties of stabilized soils; however, they are labor-intensive, time-consuming, and expensive. In this work, four artificial intelligence based models (OMC-EM, MDD-EM, UCS-EM+, and UCS-EM−) to predict the optimum moisture content (OMC), maximum dry density (MDD), and unconfined compressive strength (UCS) are developed. Experimental data covering a wide range of stabilized soils were collected from previously published works. The OMC-EM, MDD-EM, and UCS-EM− models employed seven features that describe the proportion and types of stabilized soils, Atterberg limits, and classification groups of soils. The UCS-EM+ model, besides the seven features, employs two more features describing the compaction properties (OMC and MDD). An optimizable ensemble method is used to fit the data. The model evaluation confirms that the developed three models (OMC-EM, MDD-EM, and UCS-EM+) perform reasonably well. The weak performance of UCS-EM− model validates that the features OMC and MDD have substantial significance in predicting the UCS. The performance comparison of all the developed ensemble models with the artificial neural network ones confirmed the prediction superiority of the ensemble models. Introduction The use of natural soils for habitat development is not a new concept. As of yet, it has continued to be a topic of interest to equilibrate the imbalance between the use of natural resources for human settlement development and the excessive exploitation of nonrenewable basic industrial raw materials. The industrial production of building materials, in the long run, has become a frontier of environmental calamity, be it in natural resource degradation or excessive use of energy coupled with incalculable CO 2 emission [1,2]. That being its ecofactor, the economical aspect of the products is also mostly targeting a minute segment of the global population that can afford to expend to that scale at the dearly living cost of the majority. Offsetting such disparities requires a focus on relevant researches to ensure global welfare and improved human livelihood sustainability in the long run. The use of stabilized natural soils as a sustainable alternative construction material could offer important economic and environmental benefits to society. Soil stabilization is the alteration of physicomechanical properties of soils to meet specific engineering requirements. There are different methods that can be selected for soil improvement such as chemical, dynamical, hydraulic, physical, and mechanical methods [3]. Chemical soil stabilization is the most-used technique that involves the addition of minerals to the natural soil such as lime, cement, silica fume, natural pozzolana, slag, and fly ash, or a combination of them [3,4]. The incorporated minerals chemically react with the soil constituents and result in the enhancement of strength and durability. Chemical stabilization of soil usually leads to savings in construction costs of civil engineering applications such as earth wall construction, foundation, and other earthwork purposes. Compaction is another important technique used to improve the engineering properties of earth-based construction where shrinkage can be controlled, leading to a reduction in permeability for a more stable structure [5]. The effectiveness of the compaction is usually measured by the soil's optimum moisture content (OMC) and maximum dry density (MDD). Compaction tests in the laboratory are performed to determine the OMC at which the dry densities of soils are the highest. Similar to compaction, the strength gain of soils has a paramount role in the design, construction, and long-term stability of structures of soil materials. The unconfined compressive strength (UCS) of compacted soil is usually determined by laboratory tests using an advanced machine. Though laboratory tests of OMC, MDD, and UCS are well recognized as appropriate for examining the engineering properties of stabilized soils, these tests are labor intensive, time-consuming, and costly [6]. Due to this, the proper selection of a chemical stabilizer or a combination of stabilizers and their proportion to meet natural soil, a specific engineering characteristic is a challenging activity [7]. To mitigate extensive and cumbersome laboratory testing of the OMC, MDD, and UCS of stabilized soils, it is practical to develop models that predict these values from the basic properties of natural soils and stabilizers such as plasticity, types, and quantities of stabilizers. Indeed, the properties of natural soils and stabilizers exhibit varied and uncertain behavior due to the complex and indistinct physical processes related to the formation of these materials [4]. The complexity of the behavior of the natural soil coupled with their spatial variability and the addition of stabilizers makes the development of reliable OMC, MDD, and UCS physics-based prediction models challenging. Unlike a physics-based system that performs a task by following explicit rules, such a complex problem requires intelligent systems that learn from experience. Learning a complex behavior using an artificial intelligence (AI) method is thus the best alternative. In recent years, various artificial intelligence techniques have been applied to predict OMC, MDD, and UCS of natural and stabilized soils. For example, Das et al. [8] adopted support vector machine (SVM) and three different types of ANN, which are Bayesian regularization method (BRNN), Levenberg-Marquardt algorithm (LMNN) and differential evolution algorithm (DENN) to predict the MDD and UCS of cement stabilized soil. Based on the statistical performance measure, the authors claimed that SVM outperforms ANN models. Alavi et al. [9] proposed artificial neural networks for the prediction of MDD and OMC for stabilised soil. Multilayer perceptron (MLP) architecture was adopted to develop the ANN model. The results of the tests confirm that the proposed models are satisfactory. Suman et al. [10] developed AI models to determine the MDD and the UCS of cement stabilized soil. The employed algorithms were functional networks (FN) and multivariate adaptive regression splines (MARS) and compared their performance with four models presented in [8], which are BRNN, LMNN, DENN, and SVM. Based on statistical performance measurement, the authors concluded that the adopted algorithms perform better compared to SVM and ANN based models. Bahmed et al. [3] attempted to predict OMC and MDD of lime-stabilized clayey soils using ANN. The authors claimed that the developed ANN model can be effectively utilized to predict MDD and OMC properties of stabilized clayey soils. Le et al. [11] also developed a model to predict the UCS of soils using ANN. Based on the performance evaluation, the authors concluded that ANN can accurately predict UCS. ANN and adaptive neuro fuzzy inference system (ANFIS) have been applied to predict UCS of compacted soils by Kalkan et al. [12]. They compared the performance of both models and found that the ANFIS model is superior to the ANN. Saadat and Bayat [13] also adopted ANFIS to predict the UCS of stabilized soil. The authors compared the prediction performance of ANFIS with nonlinear regression (NLR) and claimed that ANFIS outperforms NLR. Chore and Magar [14] reported that ANN can be utilized for predicting the UCS fiber-reinforced cement stabilized fly ash mixes accurately. They also compared its performance with the conventional multiple linear regression model (MLR) and concluded that ANN is superior to MLR. The earlier works' attempt to predict OMC, MDD, and UCS of stabilized soils using AI approaches are encouraging. Though several types of AI algorithms can solve complex nonlinear regression problems efficiently, most of the works mainly utilized ANN. Indeed, ANN is one of the commonly applied approaches to solve several civil engineering problems, and some practical examples can be found in [15][16][17][18]. However, ANN has several limitations. For instance, a reasonable interpretation of the overall structure of the network is often challenging, and they do not provide information about the relative importance of the predictors. ANN is also not always superior to other AI models to solve problems, and it is impossible to comprehend which algorithm would surpass for a given problem all the time. As a show case, the work of Das et al. [8] demonstrated that SVM outperforms ANN models in predicting the MDD and UCS of cement-stabilized soil. In addition, the majority of the previous works employed a low number of data obtained mainly from a single experimental study. In this work, models based on an ensemble of regression trees are developed to predict OMC, MDD, and UCS of stabilized soils. The main contribution of this work is the development of OMC, MDD, and UCS prediction models using ensemble methods and employing a wide range of stabilized soils acquired from different countries around the globe. The structure of the remainder of the paper is as follows: In Section 2, the fundamental principle about regression tree ensembles is presented. In this section, the most commonly applied methods to form ensembles of regression trees are discussed in detail. In Section 3, the applied materials and methods are elaborated. As a data-driven method, the utilized data and modeling approach are presented in detail. The results and discussion of the findings are discussed in Section 4. In this section, the performance comparison of all the developed ensemble models with the artificial neural network ones is also presented. Finally, conclusions are provided in Section 5. Regression Tree Ensembles The fundamental principle of any ensemble method is to establish a powerful predictive model by aggregating multiple base models, each of which solves the same problem [19][20][21][22][23]. Though there are several ensemble models in the literature, there are two models that use regression tree learners as a base model and have proven to be powerful in solving complex regression problems for a wide range of datasets. These ensemble models are: bagging and boosting regression trees [23][24][25]. Unlike ANN, models based on the ensemble of regression trees are interpretable and have the potential to provide relative importance of the predictors. Use of regression tree models in the field of civil engineering is not new. There are several successful practical applications in this domain [26][27][28][29]. In this work, an optimizable ensemble method that selects either the bagging or boosting method to create an ensemble of regression trees is adopted. Bagging Regression Tree In bagging regression tree, the base models are formed using multiple randomly drawn bootstrapped samples from the original dataset. This procedure is conducted numerous times until a substantial subset of training datasets are made, and the same samples can be gathered more than one time. On average, each formed bootstrapped training dataset holds N 1 − 1 e ≈ 0.63N instances, where N is the total number of samples in the original dataset. The left-out instances are known as out-of-bag observations, and it is used to assess the performance of the model. The final output of the bagging regression tree model is the average of the predicted output of the individual base models, which in turn is diminishing its variance and produces better stability [21,22,24,30]. In bagging regression tree, the base model fits the training dataset D = {(x 1 , y 1 ), (x 2 , y 2 ), . . . , (x N , y N )}, attaining the tree's predictionf (x) at input vector x. Bagging averages this prediction over a collection of bootstrap samples. For each bootstrap sample D * t , t = 1, 2, . . . , T, the model delivers predictionf * t (x). The bagged estimate is the mean prediction at x from T trees as presented by Equation (1).f (1) Boosting Regression Tree Boosting can be characterized as an enhancement of bagging that involves multiple base models by shifting the focus toward cases that encounter challenges in performing well [24,25,30]. In contrast to bagging, boosting regression trees serially construct simple tree models with improvement from one tree model to the other and fuse them to boost the model performance. Each tree is grown from a training dataset D = {(x 1 , y 1 ), (x 2 , y 2 ), . . . , (x N , y N )}, utilizing knowledge from previously grown trees. A relevant algorithm is applied to fit the training datasets D (t) , t = 1, 2, . . . , T, employing a sequence of varying weights w (1) , w (2) , . . . , w (T) , returning the tree's predictionŝ . . ,f (T) (x) for each input vector x and their corresponding weight vector w. The weight vector is usually originated by implementing an initial weight w (1) and continuously adjusted in each created base model depending on perceived errors. The weight is increased for cases in which the base model generates big errors and reduced for situations in which the model produces low errors. The final output of the models is a weighted sum of the individual model outputs, with the weights being larger for the better models. One of the best-known boosting approaches is the LSBoost (least-squares boosting) algorithm, which is adopted in this work. It begins from the null model with residuals Then, it fits a decision tree to the residuals from the model instead of the outcome y. Sequentially, the algorithm updates the residuals by adding the newly generated decision tree into the fitted function. Each of these trees can be small to a certain extent by controlling the parameter d (number of splits) in the algorithm. By fitting small trees to the residuals, thef is slowly boosted in areas where the performance is weak. The shrinkage parameter (learning rate) λ slows down the learning process further. For a small value of λ, the iteration number needed to attain a certain training error increases. The output of the boosted model is presented by Equation (2).f Raw Data To establish a reliable database for the development of OMC, MDD, and UCS prediction models, experimental data of natural and stabilized soils were collected from the published literature with certain selection criteria. This is because a data-driven model is not a common practice in this research area and there is no readily available large enough dataset. To acquire experimental data, databases of Scopus and Web of Science were largely used. Both are abstract and citation databases of the peer-reviewed literature, delivering a complete citation search by giving access to numerous databases. A set of queries that comprise the manuscript's title, topic, abstract, and keywords were carried out on every database to choose works reporting "stabilized soils". Any duplicated records from the databases were removed and then manually checked to select suitable ones. A total of 79 scientific papers which have performed experiments in stabilized soil were found and used to gather the data. Indeed, gathering data from the works of the literature was not a straightforward activity. The typical challenges were (i) the use of different measuring units, (ii) some results presented in the form of charts, and (iii) the use of different soil Appl. Sci. 2021, 11, 7503 5 of 15 classification methods. Translating all the data in preferred units and format to obtain well-sounded data required great attention and was a time-demanding task. Data consisting of 408 observations and 13 features that comprise information regarding the quantity of soils and stabilizers (cement, lime, pozzolans, and fly ash), Atterberg limits (LL, PL, and PI), compaction properties (OMC and MDD), soil classification, and unconfined compressive strength of soils at different ages were collected. The dataset covers a wide range of soils from 12 countries in Africa, America, the Middle East, South Asia, and Oceania. Modelling Process The workflow of the OMC, MDD, and UCS prediction models is illustrated in Figure 1. The initial task is retrieving the gathered experimental data that comprise information regarding the proportion of soils and stabilizers, Atterberg limits, compaction properties, soil classification, curing ages, and unconfined compressive strength of the stabilized soils. Then, data preparation follows, in which data preprocessing, feature engineering, feature selection, and scaling are performed. The next major step is to fit the data using the ensemble method (either bagging or boosting regression trees). The performance of the models is then evaluated using a previously unseen dataset. The model training and validation process is iterated by adjusting the hyperparameters till the best result is obtained. The detail of the major activities is discussed below. was not a straightforward activity. The typical challenges were (i) the use of different measuring units, (ii) some results presented in the form of charts, and (iii) the use of different soil classification methods. Translating all the data in preferred units and format to obtain well-sounded data required great attention and was a time-demanding task. Data consisting of 408 observations and 13 features that comprise information regarding the quantity of soils and stabilizers (cement, lime, pozzolans, and fly ash), Atterberg limits (LL, PL, and PI), compaction properties (OMC and MDD), soil classification, and unconfined compressive strength of soils at different ages were collected. The dataset covers a wide range of soils from 12 countries in Africa, America, the Middle East, South Asia, and Oceania. Modelling Process The workflow of the OMC, MDD, and UCS prediction models is illustrated in Figure 1. The initial task is retrieving the gathered experimental data that comprise information regarding the proportion of soils and stabilizers, Atterberg limits, compaction properties, soil classification, curing ages, and unconfined compressive strength of the stabilized soils. Then, data preparation follows, in which data preprocessing, feature engineering, feature selection, and scaling are performed. The next major step is to fit the data using the ensemble method (either bagging or boosting regression trees). The performance of the models is then evaluated using a previously unseen dataset. The model training and validation process is iterated by adjusting the hyperparameters till the best result is obtained. The detail of the major activities is discussed below. Data Preparation Data Preprocessing and Feature Engineering In the modeling process, feature engineering is the process of representing the data appropriately. It is one of the key components as it considerably influences the performance of a model. No AI algorithm is able to predict data for which it has no appropriate information. Feature engineering is often performed by the domain expert, and most of the required feature engineering activities have already been carried during the collection of the data from previously published works. Here, after careful examination of the distribution of each feature, three features (pozzolans, fly ash, and curing age) which do not have representative enough data are excluded from the database. For instance, fly ash and natural pozzolans are utilized only in 36 and 48 cases, respectively. This is a very small Data Preparation Data Preprocessing and Feature Engineering In the modeling process, feature engineering is the process of representing the data appropriately. It is one of the key components as it considerably influences the performance of a model. No AI algorithm is able to predict data for which it has no appropriate information. Feature engineering is often performed by the domain expert, and most of the required feature engineering activities have already been carried during the collection of the data from previously published works. Here, after careful examination of the distribution of each feature, three features (pozzolans, fly ash, and curing age) which do not have representative enough data are excluded from the database. For instance, fly ash and natural pozzolans are utilized only in 36 and 48 cases, respectively. This is a very small number of cases compared with other types of stabilizers. For comparison, the number of observations in the case of cement and lime is 153 and 250, respectively. Thus, retaining pozzolans and fly ash in the database causes data imbalance, which ultimately affects the performance of the model. Similarly, there are unconfined compressive strength tests performed at the age of 1, 7, 14, 28, 32, 64, and 90 days in the database, but except UCS carried out at the age of 28 days, the other age groups represent only between 1% and 15%. Thus, unconfined compressive strength tests carried out only at the age of 28 days are considered. Finally, nine features with 190 instances are selected. The data are presented in Table S1 in Supplementary Materials, and description of the features is presented in Table 1. Table 2. Feature Selection and Scaling Feature selection is the process of selecting the most relevant features from the data. This is because some features may be highly correlated and thus redundant to a certain degree or even they may be irrelevant. Generally, feature selection methods are categorized into three groups: filter, wrapper, and embedded [31,32]. The filter method is independent of the learning algorithms and relies only on the inherent nature of the data. The wrapper method demands a prespecified algorithm and is based on the utilized algorithm. The performance of each feature is adopted as the criterion for defining the final subset of features. This technique, compared to the filter, is computationally expensive but produces better accuracy. The embedded method includes the feature selection process as a component of model development. This approach is computationally inexpensive and improves the prediction performance of the predictors. The adopted ensemble method performs embedded feature selection so that it internally selects relevant features, thus enhancing the prediction performance of the model. Numerous AI algorithms demand that the selected features are on the same scale for optimal performance, which is usually achieved by transforming the features in the range [0, 1] or a standard normal distribution with zero mean and unit variance. However, the adopted ensemble methods are based on regression trees, which do not require scaling. Model Training The data were divided into training and test subsets that represented 80% and 20% of the data, respectively. The training dataset is used to fit the predictors, whereas test dataset is applied to evaluate the predictive performance of the fitted or developed model. In ensemble method, one way to achieve differences between base models is to train each model on a different subset of the available training data. Models are trained on different subsets of the training data naturally through the use of resampling methods such as cross-validation and the bootstrap, which are designed to estimate the average performance of the model generally on unseen data. Though bagging regression tree forms the training and validation set based on the embedded sampling procedure, crossvalidation resampling technique is applied for both methods. There are diverse types of crossvalidation procedures. In the case of a limited dataset, K-fold crossvalidation technique is the most desirable choice to achieve an unbiased prediction, which in turn enhances the generalization ability of the model without overfitting [21]. In K-fold cross-validation, the training data are arbitrarily partitioned into K subsets with approximately the same sizes. Each of the K subsets is employed as a validation dataset for assessing the performance of the model and the remaining (K − 1) subsets as a training dataset. In total, K models are fit, and K validation statistics are obtained. The performance evaluations from the K-folds are averaged to measure the overall performance of the model. In this work, 10-fold crossvalidation was applied. Regression trees are fitted on bootstrap samples and aggregated to create bagging regression trees. Boosting regression trees are also formed by fitting multiple regression trees iteratively in such a way that the model training at a given step depends on the models fitted at the previous steps. Every new model focuses its efforts on the most problematic instances, ultimately forming a strong learner. Using optimizable ensemble method, either bagging or boosting regression trees, four models to predict compaction and strength of stabilized soils are developed. Seven features under the category of amended soils, Atterberg limits, and soil classification are employed to predict OMC and MDD of soils. Based on the input variables, two UCS prediction models (UCS-EM + and UCS-EM − ) are developed. UCS-EM + , besides the seven features, employed two more features describing the compaction properties (OMC and MDD), whereas UCS-EM − employed the same features considered in the OMC and MDD models. Classification of the models is presented in Table 3. All the variables under the category of amended soils (soil, cement, and lime), Atterberg limits (LL, PL, and PI), soil classification, and the compaction properties (OMC and MDD). UCS-EM − 7 All the variables under the category of amended soils (soil, cement, and lime), Atterberg limits (LL, PL, and PI), and soil classification. Tuning the hyperparameters of any AI-based models is essential to optimize their performance. In the present work, with the combination of 10-fold crossvalidation, hyperparameter tuning is carried out using Bayesian optimization to improve the performance of the adopted models by finding the optimal combination of hyperparameter values that minimizes the loss function (mean square error). The Model Evaluation Once the optimum hyperparameters have been obtained for each model, their performance is assessed by measuring errors of mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE) on test dataset. MSE is the average of the squared difference between the actual and the predicted value. It is the most commonly applied loss function for regression models. MSE is calculated using Equation (3). RMSE is just the square root of the MSE. It has the same unit as the target variable. Irregularly, RMSE is preferable than MSE because understanding the error values of MSE is tricky due to the squaring effect, particularly if the target variable describes quantities in a unit of measurements. The formula of RMSE is described by Equation (4). The MAE, also known as the absolute loss, is the mean of the absolute errors (the deviation between the actual and the predicted value). Similar to RMSE, MAE is measured in the same units as the target variable. It is mathematically denoted by Equation (5). Coefficient of determination, R 2 , is another valuable quantitative measure of goodness of a prediction for a regression model. It is the fraction of response variance that is achieved by the model, and it can be described as a standardized version of the MSE, for greater interpretability of the model's performance. The value of R 2 can be computed by the mathematical relation expressed by Equation (6). The value of R 2 is bounded between 0 and 1 in case of model training, but it can become negative in case of testing. If R 2 = 1, the model fits the data perfectly with a corresponding MSE = 0. (6) where n is the number of observations, y i is the actual target value,ŷ i is the predicted output value, y the mean value of the actual target, and Var is the variance of the target variable. Results and Discussion In this section, the performance of the developed OMC, MDD, and UCS models is presented. Examining the generalization performance, how well the models can make predictions for data that was not observed during training, is very essential compared to how well they fit the training set. The coefficient of determination, R 2 , is one of the statistical measures applied to examine the generalization performance of all the models. The R 2 score indicates how well the developed OMC, MDD, and UCS models explain and predict future outcomes. It yields a score between 0 and 1. The training performance of all models is illustrated in Figure 2. Regression plot for predicted vs. measured OMC, MDD, and UCSs with their corresponding validation R-square scores, showing the true versus the predicted response. It can be perceived from Figure 2 that the score of OMC-EM, MDD-EM, and UCS-EM + models exceed 0.50. Observably, OMC-EM performs best (R 2 = 0.76), followed by UCS-EM + (R 2 = 0.69), and MDD-EM (R 2 = 0.59). This validates that these models reasonably track their corresponding target features during the training stage. Among all, UCS-EM − perform least with (R 2 = 0.49). This model entails all features under the category of amended soils (soil, cement, and lime), Atterberg (LL, PL, and PI), and soil classification, but does not entail features that describe the compaction properties (OMC and MDD). Hence, the weak performance of UCS-EM − model corroborates that the features OMC and MDD have significant importance in predicting UCS. It can be noticed that all the models exhibit a slight tendency to underestimate or overestimate their corresponding target variables in which the number of observations is very limited. For example, measurements of OMC > 25% and UCS > 3500 kN/m 2 represent only about 12% and 13% of the total observation, respectively. This insufficient number of observations in the data might be the cause for the underestimation in those ranges. This is a normal phenomenon as the algorithm did not obtain adequate observations to learn and generalize. Incorporating more observations could enhance the performance of the models. Scores of other statistical performance indicators (RMSE, MSE, and MAE) on the test dataset for all the developed models are given in Table 4. Indeed, it is impossible to compare one model with another by considering these indicators, except UCS-EM + and UCS-EM − . This is because they are predicting different variables. The lower the statistical errors are the superior the performance of the model. The RMSE, MSE, and MAE of UCS-EM + model are considerably lower than the UCS-EM − , confirming its superiority. It can also be noticed from MAE results that the average prediction errors of OMC-EM, MDD-EM, UCS-EM + , and UCS-EM − are 2.68%, 110.06 kg/m 3 , 472.33 kN/m 2 , and 622.97 kN/m 2 , respectively. This confirms that the three ensemble models (OMC-EM, MDD-EM, and UCS-EM + ) performed rationally well on previously unseen data, considering the fact that their corresponding median values are 11.30%, 1820 kg/m 3 , and 2260 kN/m 2 . All the results are valid only for the employed dataset. The performance of each model could be enhanced if more data are utilized. The optimized hyperparameters are also presented in the Table 4. As presented in Section 3.2.2, five hyperparameters were considered to optimize the performance of the models. Bagged regression trees yielded optimal performance for all models. The minimum leaf size, number of learners, and number of predictors for the samples of each model are mostly different. Appl. Sci. 2021, 11, x FOR PEER REVIEW 10 of 15 Scores of other statistical performance indicators (RMSE, MSE, and MAE) on the test dataset for all the developed models are given in Table 4. Indeed, it is impossible to compare one model with another by considering these indicators, except UCS-EM + and UCS-EM − . This is because they are predicting different variables. The lower the statistical errors are the superior the performance of the model. The To compare the performance of the developed four ensemble models, four other models employing ANN algorithms are developed. These models are OMC-ANN, MDD-ANN, UCS-ANN + , and UCS-ANN − . All the ANN models have three layers: an input, a hidden, and an output layer. The number of the input neurons for each model corresponds to the number of the predictors, which is seven neurons for OMC-ANN, MDD-ANN, and UCS-ANN − and nine neurons for UCS-ANN + . The number of neurons utilized in the hidden layer was ten, which was determined based on the generalization error after performing several trainings. The number of neurons at the output layer is one for all models, which is the feature going to be predicted. The data were also randomly divided into three clusters: training, validation, and test datasets, which hold 70%, 15%, and 15% of the dataset, respectively. Once the datasets are ready, the network was trained using a Levenberg-Marquardt algorithm. The difference between the actual and the predicted OMC, MDD, and UCSs by the developed four ANN models are calculated, and their distributions are visualized with a boxplot in Figure 3. The median of the errors is designated by a red line within the blue box, embracing the middle 50% (25th-75th percentiles) of the errors. It can be seen from Figure 3 that the medians of errors of all the models are closer to either the first or the third quartile. For instance, the medians of the errors of OMC-ANN and UCS-ANN + models are closer to the first quartile. This means that the distributions of the errors are slightly skewed to the right. The whiskers stretch from the ends of the box to the smallest and largest error values. Errors greater than 1.5 box lengths above the whiskers are outliers and characterized by a red plus sign. It can be observed that there are a significant number of outliers in all models. Statistical performance indicators (MSE and R 2 ) of all the ensemble and ANN models are given in Table 5. The lower the MSE and the higher the R 2 are, the more superior the model. It can be observed from Table 5 that the MSE errors of all models which employed ensemble of regression trees are lower than the corresponding models based on ANN. For instance, the MSE of the OMC-EM compared to OMC-ANN is lower by 34%. The R 2 of all the ensemble models is higher than ANN models. For example, the R 2 of OMC-EM is higher than OMC-ANN by 31%. All these confirm that the ensemble methods outperformed the ANN models on the utilized dataset. The developed ensemble models are a promising approach to evaluate the performance of compaction and strength properties of stabilized soils. Housing blocks based on stabilized soils have been already gaining recognition in many emerging countries, especially in the global south as the weather condition favors their wide application. Indeed, the stabilizers type varies depending on their availability. For instance, a recent research work by Admassu K. [6,7] demonstrated the possibility of stabilizing soils using locally available raw lime and raw natural pozzolan on three types of soils. Across the considered soil range, varying improvement effects were recorded in OMC, MDD, and UCS. In general, the soils were effectively stabilized to certainly induce physical and mechanical property changes on the main ingredients, making them fit for the production of wall building blocks and jointing mortar. The trend shades promising limelight on the initiated make for a futuristic affordable, sustainable, and ecofriendly alternative earth-based built environment. With the availability of more and more such experiments (data), the same models with new training dataset could be utilized to predict the physical and mechanical properties of stabilized soils. The developed ensemble models are a promising approach to evaluate the performance of compaction and strength properties of stabilized soils. Housing blocks based on stabilized soils have been already gaining recognition in many emerging countries, especially in the global south as the weather condition favors their wide application. Indeed, the stabilizers type varies depending on their availability. For instance, a recent research work by Admassu K. [6,7] demonstrated the possibility of stabilizing soils using locally available raw lime and raw natural pozzolan on three types of soils. Across the considered soil range, varying improvement effects were recorded in OMC, MDD, and UCS. In general, the soils were effectively stabilized to certainly induce physical and mechanical property changes on the main ingredients, making them fit for the production of wall building Apart from that, incorporating additional features which describe the soil and the stabilizers such as soil grading, soil compaction state, or describing stabilizers explicitly (e.g., type of cement and lime) could enhance the prediction capability of the models. Both compaction and grading play a significant role in keeping the integrity of the stabilized soil. The type and proportion of the stabilizers content also significantly influence the development of the unconfined compressive strength of the stabilized soil. The inclusion of experimental data in the database in which the proportion of the utilized stabilizers in each range such as 0-10%, 10-20%, and 20-30% is sufficient enough. The models can be adopted to formulate the proportion of stabilizers that could meet the desired compaction properties (OMC and MDD) or strength of stabilized soils. It is important to note that the prediction accuracy of the current model is not high enough due to limited data. With sufficient data (both observations and number of predictors), the model performance could be enhanced, and it can substitute the cumbersome and expensive laboratory tests of stabilized soils. Moreover, it could assist the scientific community to obtain a better insight into how different complex soil types and stabilizers in relative proportions could be estimated to predict the possibly achievable compaction and strength properties of stabilized soils. Conclusions A total of four artificial intelligence based models to predict the compression strength properties of stabilized soils were developed. These are OMC-EM, MDD-EM, UCS-EM + , and UCS-EM − . An optimizable ensemble method which selects either bagging or boosting regression trees to create an ensemble was adopted. To establish a reliable database for the development of the models, experimental data of a wide range of stabilized soils were collected from previously published works. The OMC-EM, MDD-EM, and UCS-EM − models employed seven features which describe the proportion and types of stabilized soils (soil, cement, and lime), Atterberg limits (LL, PL, and PI) and classification group of soils. The UCS-EM + model utilized a total of nine input features. These are all the seven features as well as the OMC and MDD properties of the stabilized soils. The performance of all the developed models demonstrated their promising application in the prediction of OMC, MDD, and UCS. The weak performance of UCS-EM − model corroborates that the features OMC and MDD play a significant role in predicting UCS. The performances of all the developed ensemble models were compared with the artificial neural network based models. The comparison corroborated that the ensemble models outperform the ANN models. The performance of all the models could be improved further with more data and can be applied to determine the optimal proportion of stabilizers that could meet the desired OMC, MDD, and UCS.	2021-09-27T18:55:03.618Z	2021-08-02T00:00:00.000	{ "year": 2021, "sha1": "8fd738a30b959643f8e021473069d986cce10a34", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2076-3417/11/16/7503/pdf?version=1629867211", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "826646202a0df2538d56e1f4d8cf704fd732d55a", "s2fieldsofstudy": [ "Geology" ], "extfieldsofstudy": [ "Environmental Science" ] }
253041139	pes2o/s2orc	v3-fos-license	Effect of Blood Transfusion on Cerebral Hemodynamics and Vascular Topology Described by Computational Fluid Dynamics in Sickle Cell Disease Patients The main objective of this study was to demonstrate that computational fluid dynamics (CFD) modeling can be used to study the contribution of covert and overt vascular architecture to the risk for cerebrovascular disease in sickle cell disease (SCD) and to determine the mechanisms of response to therapy such as chronic red blood cell (cRBC) transfusions. We analyzed baseline (screening), pre-randomization and study exit magnetic resonance angiogram (MRA) images from 10 (5 each from the transfusion and observation arms) pediatric sickle SCD participants in the silent cerebral infarct transfusion (SIT) trial using CFD modeling. We reconstructed the intracranial portion of the internal carotid artery and branches and extracted the geometry using 3D Slicer. We cut specific portions of the large intracranial artery to include segments of the internal carotid, middle, anterior, and posterior cerebral arteries such that the vessel segment analyzed extended from the intracranial beginning of the internal carotid artery up to immediately after (~0.25 inches) the middle cerebral artery branching point. Cut models were imported into Ansys 2021R2/2022R1 and laminar and time-dependent flow simulation was performed. Change in time averaged mean velocity, wall shear stress, and vessel tortuosity were compared between the observation and cRBC arms. We did not observe a correlation between time averaged mean velocity (TAMV) and mean transcranial Doppler (TCD) velocity at study entry. There was also no difference in change in time average mean velocity, wall shear stress (WSS), and vessel tortuosity between the observation and cRBC transfusion arms. WSS and TAMV were abnormal for 2 (developed TIA) out of the 3 participants (one participant had silent cerebral infarctions) that developed neurovascular outcomes. CFD approaches allow for the evaluation of vascular topology and hemodynamics in SCD using MRA images. In this proof of principle study, we show that CFD could be a useful tool and we intend to carry out future studies with a larger sample to enable more robust conclusions. Introduction Sickle cell disease (SCD) is the most common inherited hemoglobinopathy worldwide, affecting over 300,000 live births each year [1]. SCD results from a base substitution at the sixth amino acid position in the β-globin chain [2], which causes red blood cells (RBCs) to "sickle" under hypoxic and/or acidotic conditions, which results in microvascular occlusion, infarction, and end organ damage. SCD causes significant morbidity and early mortality; neurovascular complications are particularly devastating and range from overt stroke (clinical stroke) to progressive cognitive decline even in the absence of neuroanatomical changes [3]. Indeed, 11% of untreated children with SCD will experience an overt ischemic stroke by age 20 years, while approximately 30% of individuals with SCD have evidence of silent cerebral infarctions (SCI), defined as areas of white matter hyperintensity seen on T2-weighted brain MRI [4,5]. These "silent" infarctions are not truly silent, as they are associated with worse performance on measures of cognitive function (using the proxy measure of full-scale intelligence quotient) compared to individuals with SCD without SCI [6]. The large-vessel disease of overt stroke in SCD has been well-characterized, with evidence of stenosis, downstream occlusion, and tortuosity of the affected vessels, predominantly within the Circle of Willis. These changes correlate with an elevated arterial blood flow velocity of greater than 200 cm/s (cm/s) in the anterior cerebral artery, middle cerebral artery, or internal carotid artery on transcranial Doppler ultrasound (TCD) [7][8][9]. SCI has been associated with intracranial large vessel stenosis, as well as extracranial internal carotid artery stenosis and significant anemia (baseline hemoglobin below 7 g/dL) [10]. Additionally, evidence from angiography and autopsy have documented several pathological changes in the cerebral macro-and microvasculature including stenosis and fibrosis, exuberant intimal growth and endothelial proliferation, and formation of sickle red cell sludge in small blood vessels as contributors to SCD-related cerebral vasculopathy [11][12][13][14]. Studies in subjects without SCD using endothelialized microfluidics devices have shown that computational fluid dynamics (CFD) models built from magnetic resonance angiography (MRA) images can be used to identify sections of a vessel or flow channel with uneven internal arterial surfaces that may induce regions of low wall shear stress, which are associated with greater endothelial activation and possibly intimal hyperplasia that may predispose to stenosis [15][16][17][18]. In a recent study with three patients (one healthy control and 2 with SCD), Rivera et al. applied CFD to the internal carotid artery (ICA) and its main branches, and demonstrated the presence of internal arterial wall surfaces with regions of low wall shear stress and more disturbances in blood flow; they hypothesized that these regions are predisposed to stenosis and possibly contribute to the observed higher TCD velocities and stroke risk in the children with SCD [19]. However, SCD patients without evidence of extra-and intracranial stenosis and normal TCD velocities still experience overt stroke and SCI [20][21][22]; further investigation into the molecular and structural mechanisms for these pathologies is warranted. In this study, we used MRA images from children with SCD with baseline TCD velocities <200 cm/s to create CFD models to characterize the topology and flow parameters of the left and right ICA and main branches in ten patients, five each from the observation and transfusion arms of the SIT trial. We hypothesized that velocity profile, wall shear stress, and vessel topology (tortuosity) are correlated and can be used as markers of progression of cerebrovascular disease and response to blood transfusion therapy in children with SCD, especially in the setting of TCD velocities <200 cm/s. • This study will represent the largest study to date to utilize MRI images from patients with SCD to model blood flow and wall shear stress in segments of the ICA and MCA. • It will also be the first to perform this modeling in a longitudinal fashion, in children with normal TCD velocity at baseline thus allowing us to take the step towards determining whether the CFD measures could have predictive benefit. • It is also the largest study to date to use MRA images to model hemodynamic behavior and wall shear stress in individuals with SCD. Materials and Methods To aid the reader, we have provided Supplementary Figure S1, which summarizes the overall workflow and is thus a super summary of the methodology for this study. Furthermore, because this is no longer considered human subject research, a prior IRB approval was not required. SIT Trial Overview The SIT trial was a multicenter randomized clinical trial to determine whether chronic RBC transfusion was efficacious at preventing the development of new SCI and/or progression of existing SCI over a three-year period in children with SCD and TCD velocities of <200 cm/s [21,23]. Briefly, the study included children ages 5-15 years with SCD, specifically the higher risk genotypes of HbSS and HbSβ 0 thalassemia, a normal or conditional screening TCD (defined as time-averaged maximum mean velocity of <200 cm/s in the anterior cerebral artery (ACA), middle cerebral artery (MCA), and internal carotid artery (ICA)), and at least one infarct-like lesion on the screening MRI scan; children with other SCD genotypes and those already on disease-modifying therapy with either hydroxyurea or chronic transfusion therapy were excluded. SCI was defined as an area of hyperintensity at least 3 mm in one dimension, visible in at least 2 planes on a FLAIR T2-weighted MRI sequence, and not associated with a clinical neurological change [21,23]. Children were first screened to ensure that they met the eligibility criteria and then they were randomized to either chronic (monthly) red blood cell (cRBC) transfusions or observation only (the standard-of-care) for 36 months; participants underwent screening MRI/MRA, MRI/MRA at baseline (i.e., prior to randomization), and MRI/MRA after 36 months of cRBC therapy or observation [23]. The SIT trial utilized a standardized MRI protocol described in Supplementary Appendices by Casella et al. [23]. By utilizing participants which were imaged on different MRI scanners, though using a standardized MRI sequencing protocol, we strove to increase the generalizability of our findings. Because the original SIT trial relied on MRI findings as a primary endpoint for the study, the imaging data had to be very consistent across MRI scanners. Case Selection for This Study We randomly selected the MRA images from 10 participants (5 in the observation arm and 5 from the transfusion arm) and included images from all three time points (screening, pre-randomization, and 36 months), when available, to perform CFD and vessel tortuosity analysis. The members of our study team (SP, EG and RB) who completed the analysis were blinded to each participant's treatment status. We also obtained information on each participant's age at baseline, TCD velocity recorded at each time point or at a date closest to the date of MRI/MRA acquisition (when available), presence/development of neurological events (i.e., new SCI, transient ischemic attack (TIA), or overt stroke) during the three years of study follow up. 3-D Model Development The 3-dimensional models of segments of the internal carotid artery and its main branches were generated from raw MRA data using 3D Slicer [24] and then segmented into left and right sides using Autodesk Meshmixer (Autodesk, San Rafael, CA, USA). The ICA vessel was cut immediately after (~0.25 inches) the emergence of its major branches (MCA, ACA and PCA) before its branching. The result was a stereolithographic (STL) 3D image/model of the vessel segment, which was further processed and smoothed in Fluent SpaceClaim before moving onto the CFD preprocessing. The 3D models were imported into Ansys Student (Ansys Inc., Cannonsburg, PA, USA) 2021R2 (Workbench-Fluent) for 3-dimensional time-dependent CFD analysis using laminar flow assumption. A more detailed description of the CFD processing and analysis is presented below. Mesh Generation The finite element method (FEA) was used to determine correlation between the vessel geometry and hemodynamic parameters. The inlets and outlets were defined for the 3D vessel geometry as ICA and MCA/ACA respectively. The vessel geometry was discretized into a polyhedral mesh for better gradients approximation [25]. Grid/mesh convergence study was done to make sure that the velocity results are independent of the mesh size. A table for the variation of velocity and WSS for a varied number of elements is shown in Supplementary Table S1. A wide range of number of elements, i.e., from 6500 to 18,000 was considered and the effect of elements number on velocity and wall shear convergence was analyzed. The table shows that the change in velocity and shear values was low compared to the increase in mesh element number and computational time. Thus, to keep the percent error of variation for velocity and WSS at less than 5%, the mesh was set at medium smoothing with approximately 9500 elements to optimize the simulation time while maintaining the accuracy of velocity and shear parameters. The simulation result showed convergence as shown in the Supplementary Figure S2. The time-averaged values for wall shear and velocity were calculated once both parameters were stable, and change was less than 3%. The line graph for velocity and wall shear stress along the vessel wall for one patient (AAF) in the observation arm and one patient (HAE) in the cRBC transfusion arm is shown in Supplemental Figure S3. where µ∇ 2 u is the reduced form of shear stress divergence , D/Dt is material derivative, u = flow velocity, ∇ = divergence, p = pressure, µ = dynamic viscosity, and ∇ 2 = Laplace operator. ∇.τ as ∇.u = 0 for incompressible fluid. Blood in these simulations was treated as a Newtonian fluid with constant viscosity (µ = 0.03 g/(cm·s), ρ = 1.06 g/cm 3 ). Both the artery wall and blood (1.22 g/cm 3 ) were assumed healthy (i.e., physiological levels of blood cells and other blood components and no evidence of vessel wall leakage). The boundary conditions were defined with a constant inlet velocity of 130 cm/s (which is based on the lower limit of the normal reported by Adams et al.) [7,26,27], with a null pressure gauge at the outlet. The velocity along the artery walls is assumed to be zero (no-slip condition). Solution The area-weighted velocity at the MCA outlet and shear throughout the vessel wall were calculated at each time step. Here, the shear stress was estimated from the following equation, Here, the latter term is normal components of shear stress projected as tangential form. The clinical significance of wall shear in boundary wall is well defined [19,28]. The residual error convergence levels were set to 10 −6 for each equation. We performed laminar transient simulations of time steps of 0.001 s with the total time of 3 s.. Result The time averaged mean velocity (TAMV) was calculated by integrating the area under the curve (AUC) after the velocity had stabilized in the simulation, and this velocity estimate was captured in Equation (1) below. The TAMV was calculated once the solution was stabilized over the multiple time steps where the changes in velocity and wall shear stress values were Brain Sci. 2022, 12, 1402 5 of 14 less than 3% between successive trials as well as between maximum and minimum value considered. Both values in the final time step were excluded from the calculation. . The wall shear contour plots were generated at T = 3 s in the entire vessel wall and streamline plots were generated starting from the inlet. Tortuosity Index Calculation Arterial tortuosity is defined as the measure of the convoluted pathway of blood flow within a vessel compared to the direct pathway between the two ends of the vessel and has been implicated in both genetic cardiovascular disease syndromes and in the development of white matter hyperintensities in older adults [29,30]. We calculated the tortuosity index of the vessel segment from the ICA to the MCA outlet of each 3D blood vessel model using 3D Slicer with VMTK (Vascular Modeling Toolkit). For our study, arterial tortuosity was calculated using previously published approaches [31][32][33] as the ratio of the center line length (L1) to the geometric length (L2) and then subtracting 1 (which indicates a perfect vessels with no tortuosity) from the result (Tortuosity index = (L1 ÷ L2) − 1) (see Figure 1), with larger values indicating more tortuous vessels. transient simulations of time steps of 0.001 s with the total time of 3 s.. Result The time averaged mean velocity (TAMV) was calculated by integrating the area under the curve (AUC) after the velocity had stabilized in the simulation, and this velocity estimate was captured in Equation (1) below. The TAMV was calculated once the solution was stabilized over the multiple time steps where the changes in velocity and wall shear stress values were less than 3% between successive trials as well as between maximum and minimum value considered. Both values in the final time step were excluded from the calculation. . The wall shear contour plots were generated at t = 3 s in the entire vessel wall and streamline plots were generated starting from the inlet. Tortuosity Index Calculation Arterial tortuosity is defined as the measure of the convoluted pathway of blood flow within a vessel compared to the direct pathway between the two ends of the vessel and has been implicated in both genetic cardiovascular disease syndromes and in the development of white matter hyperintensities in older adults [29,30]. We calculated the tortuosity index of the vessel segment from the ICA to the MCA outlet of each 3D blood vessel model using 3D Slicer with VMTK (Vascular Modeling Toolkit). For our study, arterial tortuosity was calculated using previously published approaches [31][32][33] as the ratio of the center line length (L1) to the geometric length (L2) and then subtracting 1 (which indicates a perfect vessels with no tortuosity) from the result (Tortuosity index = (L1 ÷ L2) − 1) (see Figure 1), with larger values indicating more tortuous vessels. Statistical Analysis Statistical analysis was completed using SAS version 9.4. Demographics were compared between cRBC transfusion and observational groups. Fisher's exact test or the chisquare test was used to analyze the categorical variables. Differences in continuous variables were evaluated using Student's t-test and Wilcoxon rank-sum test. TAMV, tortuosity, and WSS were compared between the right and left ICA/MCA at the pre- Statistical Analysis Statistical analysis was completed using SAS version 9.4. Demographics were compared between cRBC transfusion and observational groups. Fisher's exact test or the chi-square test was used to analyze the categorical variables. Differences in continuous variables were evaluated using Student's t-test and Wilcoxon rank-sum test. TAMV, tortuosity, and WSS were compared between the right and left ICA/MCA at the pre-randomization time point. TAMV at pre-randomization was then subtracted from the TAMV value at the 36-month timepoint to create change in TAMV values. Mean and standard deviation of change in TAMV values was then calculated. The mean change in TAMV was compared between the cRBC and observation groups. This process was repeated for tortuosity and WSS. The right and left ICA/MCA were analyzed separately, except when comparing left to right at the pre-randomization time point. We then performed Spearman's correlation analysis to determine if there is a correlation between participant's age, TCD velocity at screening and TAMV or WSS at screening or study exit. We did not examine the other time points because of the high rate of missingness for the TCD velocity data at the pre-randomization. Demographics Of the ten patients included in our analysis, five had been randomized to cRBC and five to the observation group. There was no difference in age: 7.9 (1.5) vs. 7.9 (2.0) years between the participants in the cRBC arm compared to the observation arm. Each group was 20% female. At study entry the mean TCD velocity for the observation group was 136 cm/s (24.7) compared to 153 cm/s (21.8) in the cRBC group, which was not significantly different (p = 0.3). Among the 10 participants, there were a total of three cerebrovascular events during the 36 months of follow up: two transient ischemic attacks (one in the cRBC group and one in the observation group) and a new SCI in the observation group. None of the participants that experienced a neurological event had initial TCD velocities ≥200 cm/s. However, we observed that the participant who experienced a TIA on transfusion had a TCD velocity of 241 cm/s at study exit. This participant also had a TAMV of 272 cm/s on the left, based on CFD modeling from the pre-randomization MRA. Similarly, the participant in the observation arm who had TIA also had a pre-randomization TAMV of 303 cm/s on the left side; unfortunately there was no exit TCD velocity recorded for this participant. For both participants, the CFD-derived TAMV was normal on the right side and was 149 cm/s and 194 cm/s, respectively, on the left side, at study exit. The participant who had an SCI in the observation arm did not have any TAMV (from CFD modeling) or TCD velocity recording that was considered abnormal or even conditional, based on current clinical cutoffs (See Supplementary Table S1). Arterial Blood Flow Velocity There was no statistically significant difference between the cRBC and observation groups with regard to mean TAMV or change in TAMV from baseline, possibly due to the small sample size (Table 1). A closer look at the data (see Table 2), show a marginally significant but negative correlation (r 2 = −0.65, p = 0.06) between baseline TCD velocity in all 10 participants and TAMV on the left side, at the screening time point. Similar negative correlations were observed between baseline TCD velocity and screening TAMV on the right, and baseline TCD velocity and TAMV at study exit bilaterally; however, these correlations were not statistically significant. Notably, both TCD velocity and CFD-derived TAMVs at screening were negatively correlated with participants age at baseline. When we examined the relationship between velocities and WSS, we observed a statistically significant correlation (r 2 = 0.74, p = 0.04) between TAMV on the right and WSS on the right at the study exit. We also observed a similar (positive) trend in correlation between TAMV on the left and WSS on the left, as well as between TAMVs and WSS at baseline; however, these were not statistically significant (see Table 2). Figure 2A,B are representative velocity streamlines obtained from CFD simulations for participants randomized to the observation and transfusion arms, respectively. For each participant we also show simulation images from the pre-randomization (Pre-rand) and study exit (MRI-36) time points. Mean Change in TAMV in cm/s (SD) Prerandomization Vessel Tortuosity Comparison of vessel tortuosity between the cRBC group and the observation group (Table 3) demonstrated no statistically significant difference in degree of ICA to MCA segment tortuosity between the groups. We observed a slight decrease in vessel tortuosity on the left among the cRBC arm, while there was a slight increase in tortuosity on the same side, for the observation arm. While these observations were not statistically significant, and we note the small sample size of this study, they demonstrate proof of principle for our methods and thus the potential when applied to a larger sample. Wall Shear Stress At the screening time point there was no statistically significant correlation between WSS and either TAMV or screening TCD velocity, except for TAMV and WSS on the right. As shown in Table 4, we did not observe a statistically significant difference between the cRBC and observation group with regard to WSS at the pre-randomization time point (p = 0.09 and 0.11, on the right and left respectively). Additionally, at the study exit time point, the WSS was not significantly different between groups, either on the left or right side. The change from baseline was also not significantly different between groups (p = 0.08 and 0.30, on the right and left respectively). It is worth noting that based on currently available data, the physiologic level of WSS is 20-30 dynes/cm 2 [34] with some studies suggesting that the lower boundary could be as low as 18 dynes/cm 2 [35]. Low (9 dynes/cm 2 ) and supra low or sub-physiologic (4.5 dynes/cm 2 ) WSS levels are associated with endothelial and consequently arterial remodeling [34][35][36] and endothelial activation [37,38]. Both endothelial remodeling and activation are key components of the vascular pathobiology of SCD-related cerebrovascular complications. Thus, as shown in Table 4, at baseline, the average WSS for the cRBC arm was closer to the sub-physiologic level, while that for the observation arm was in the supraphysiologic level of ≥36 dynes/cm 2 [35]. At study exit in Table 4 the average WSS level for the cRBC arm has increased from the sub-physiologic to the supraphysiologic levels. Unlike sub-physiologic levels of WSS, supraphysiologic levels has been reported to increase the likelihood for thrombosis [34,35]. The implication of this in SCD will become clear in larger sample analysis. Looking at Supplemental Table S2, a closer examination of the WSS data for the two participants who developed TIA indicated that the participant on cRBC had WSS values of 100.1 dynes/cm 2 and 96.4 dynes/cm 2 on the right and left side, respectively, at the study exit timepoint. Similarly, the participant in the observation arm who developed TIA also had WSS values that were 79.0 dynes/cm 2 and 183.1 dynes/cm 2 on the right and left side, respectively, at the pre-randomization time point. We did not observe elevated (>30 dynes/cm 2 ) WSS values in the participant who developed SCI, however, this participant, like most in this study, had sub-physiological levels of WSS. Figure 3A,B are representative images of the distribution of WSS along the analyzed vessel segment, obtained from CFD simulations for participants randomized to the observation and transfusion arms, respectively. For each participant we also show simulation images from the pre-randomization (Pre-rand) and study exit (MRI-36) time points. In the Boult et al. [4] framework for novelty, the critical elements for defining novelty are task-dependent dissimilarity functions D w,T ;Et D o,T ;Et , and associated thresholds δw and δo for the world and observational space respectively. This paper shows that multiple dissimilarity operators support defining multiple subtypes of novelty (e.g., novel agents, novel actions by agents, novel interrelations, etc). We also define a WOW-agent agent-dissimilarity-operator and show how to use Extreme Value Theory (EVP) to deal with the inherent uncertainty in the world while defining an effective threshold δa for the WOW-agent to declare novelty subtypes. Finally, we show that EVP can be effectively used to determine probabilities and the novelty detection thresholds in a 3D CartPole environment. pre-randomization and after 36 months. In the right ICA WSS further decreased from 6.9 dynes/cm 2 pre-randomization to 5.0 dynes/cm 2 at 36 months, while in the left ICA the WSS increased from 5.0 dynes/cm 2 to 6.9 dynes/cm 2 . Panel "(B)" represents wall shear stress in a participant in the cRBC treatment arm with MRA at pre-randomization and after 36 months. In the right ICA WSS increased from 6.2 dynes/cm 2 pre-randomization to 85.5 dynes/cm 2 at 36 months, while in the left ICA the WSS increased from 5.3 to 88.3 dynes/cm 2 . cRBC = chronic red blood cell infusion, MRA = magnetic resonance angiography, WSS = wall shear stress. Discussion We observed no differences in change in TAMV, WSS, and vessel tortuosity at 36 months between SCD participants treated with cRBC transfusion and those in the observation group. At baseline and 36 months our SCD participants had higher vessel tortuosity than adult healthy controls, with comparable tortuosity indices to adults with connective tissues diseases such as Marfan's Syndrome and Loeys-Dietz Syndrome [39]. Increased vessel tortuosity has been seen in extracranial carotid and vertebral arteries in adults with SCD [40], as well as intracranially in a mouse model of SCD [41]. Similarly, the WSS was higher in our pediatric SCD participants than in healthy adult and pediatric controls in other studies using similar MRA approaches [42,43]. Our TAMV was generally higher than approaches which have used transcranial Doppler to quantify TAMV in SCD [44] which is consistent with previous findings by Rivera et al. [19]. Thus, our findings add to the literature describing intracranial vasculopathic changes resulting in greater TAMV, WSS, and vessel tortuosity in SCD. Additionally, the higher TAMV and WSS observed in our study could be attributed to the fact that we integrated our average data along the entire vessel segments (usually longer); thus, we included "hot spots" with very high local velocity and WSS, which is different from the way TCD velocity measures (TAMV) are calculated [7,45]. Among the ten participants, two in the observation group and one in the cRBC group had cerebrovascular events during the 36 months of follow up. These participants who developed cerebrovascular events had TCD velocities < 200 cm/s at baseline and prerandomization time points. The participant on cRBC transfusion had a CFD-derived TAMV of 272 cm/s on the left at the pre-randomization time point, a TCD velocity of 241 cm/s at study exit, and WSS values of 100.1 dynes/cm 2 and 96.4 dynes/cm 2 on the right and left side, respectively, at the study exit timepoint. Similarly, the participant in the observation arm who developed TIA also had a CFD-derived TAMV of 303 cm/s on the left at the pre-randomization time point and WSS of 79.0 dynes/cm 2 and 183.1 dynes/cm 2 on the right and left side, respectively, at the pre-randomization time point. The TCD velocity value for this participant was missing for the study exit time point, while the other time points were within normal limits. Thus, based on the CFD TAMV and the WSS levels, both these participants were at increased risk for stroke as previously described [7,35,46]. While Liu et al. found that large vessel vasculopathy was associated with increased white matter disease in adults with SCD [30], our study demonstrates that cerebrovascular events occur in children with SCD with even mild large vessel vasculopathy, suggesting additional pathophysiologic mechanisms which contribute to cerebrovascular events. Certainly, large vessel vasculopathy contributes to the development of silent infarcts as well as overt cerebrovascular disease, as Guilliams et al. found a greater distribution and overall density of silent cerebral infarcts in areas with large vessel vasculopathy [47]. Low cerebral blood flow may be an additional factor resulting in cerebrovascular disease in SCD [48], particularly in the setting of large vessel vasculopathy. Likely large vessel vasculopathy interacts with a variety of factors resulting in cerebrovascular disease in SCD. One question that remains unanswered is whether the vasculopathic changes occurring in large vessels in SCD are the same as those occurring in small cerebral vessels; we hypothesis similar changes occurring in the small cerebral blood vessels contribute to SCI in SCD. However, this hypothesis has yet to be tested and we hope to do so in future studies. Furthermore, also unclear is why such dramatic changes in WSS values exist, especially in the cRBC arm. Whether this is a consistent trend is one of the questions we hope to answer with a larger sample. At the moment, we can only speculate that it might be due to a maladaptive response to changes in hemoglobin levels and other blood rheological properties. However, this will need to be supported by laboratory evidence. Our study did not demonstrate changes in the WSS and vessel tortuosity in the cRBC transfusion group, as chronic transfusion therapy has been shown to improve vessel tortuosity in other studies and is one of the bases for the recommendation of cRBC transfusions in the setting of an abnormal TCD [7,8,46,49]. This may have been due to our small sample size. As well as preventing overt stroke, the SIT trial demonstrated a 58% relative risk reduction in the development of new SCI in those placed on the cRBC transfusion protocol [21]. In addition to the improvement in vessel tortuosity and wall shear stress, this is also likely due to an increase in both total hemoglobin and improved oxygen carrying capacity with hemoglobin A, as well as lower cerebral blood flow velocity, allowing for better perfusion of the watershed areas of the brain, which are particularly susceptible to SCI. Hydroxyurea is speculated to work in a similar way to cRBC transfusions by also increasing total hemoglobin and increasing oxygen-carrying capacity through an increase in fetal hemoglobin [50][51][52]. One obvious limitation of this study is the small samples size of ten SCD participants, which resulted in wide ranges in standard deviations, limiting our ability to detect true differences where they may exist. However, as stated earlier, this study is a demonstration of the feasibility of our model, with the plan to increase the sample size in future studies/analysis. Another limitation is non-uniform availability of hematological and other clinical measures such as blood oxygen levels. It is difficult to infer random missingness due to small sample size. Additionally, geometries were rigid, and we did not have patientspecific boundary conditions. Furthermore, viscosity was assumed to be constant and akin to literature values. Blood is a non-Newtonian fluid. Our simulations were for the large arteries, not the smaller vessels, where shear rates are lower and produce differences in viscosity between Newtonian and non-Newtonian models. An additional limitation was the availability of TCD data. The SIT trial concluded in 2014 and unfortunately TCD TAMV were not recorded for each vessel (MCA, ACA, ACA, PCA, and basilar), and only the highest TAMV was recorded for the participant at a given time point (screening, prerandomization, and at 36 months). Therefore, we utilized these data since they reflected the highest TAMV at a given time point. Another limitation is the focus on the ICA and its branches, without analysis of the PCA and their branches. Evaluations of intracranial stenosis and tortuosity in SCD have primarily focused on the ICA/MCA junction using direct cerebral angiography and TCD [8,53,54]. As most cerebral infarcts in SCD occur in the ICA and MCA distribution, we prioritized analysis of the ICAs over the posterior cerebral arteries in this proof of principle work. Conclusion In this study, we demonstrated that computational fluid dynamics modelling can be applied to real-world magnetic resonance angiography imaging to determine blood flow velocities and wall shear stress, particularly in indiviudals with SCD. Although small, this first proof-of-principle study has garnered valuable insight into the pathophysiology of SCD-related large vessel vasculopathy. Future studies will involve a large sample size with the goal to better define the vascular changes that predict cognitive impairment in individals with sickle cell disease. Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/brainsci12101402/s1, Figure S1: Study framework/workflow; Figure S2: Simulation convergence results; Figure S3: The line graph for velocity and wall shear stress along the vessel wall for one patient (AAF) in the observation arm and one patient (HAE) in the cRBC transfusion arm. Table S1: Mesh convergence study showing percentage variation in velocity and wall shear with respect to baseline value (9500 elements); Table S2: Presentation of some relevant individual characteristics for each participant. Institutional Review Board Statement: The original SIT trial was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of each of the 29 participating institutions. However, as the images used in the present study were de-identified and did not directly involve human subjects, IRB approval was waived/not required. Informed Consent Statement: Informed consent was obtained from all subjects upon enrollment in the SIT trial with the inclusion that de-identified imaging data could be used for future studies such as this one. Data Availability Statement: The data included in this study are available from the corresponding author upon request and processing of required DUA/MTA. Conflicts of Interest: The authors declare that the research was conducted in the absence of any relevant commercial or financial relationships that could be construed as a potential conflict of interest.	2022-10-21T15:33:20.686Z	2022-10-01T00:00:00.000	{ "year": 2022, "sha1": "d57525222900991364b36b070f46827f86cf4ed8", "oa_license": "CCBY", "oa_url": "https://doi.org/10.3390/brainsci12101402", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "9f2263b14ee6c4aa03a856b5484b84e7d40f3cd7", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
259684120	pes2o/s2orc	v3-fos-license	Fungal Endophytes: Microfactories of Novel Bioactive Compounds with Therapeutic Interventions; A Comprehensive Review on the Biotechnological Developments in the Field of Fungal Endophytic Biology over the Last Decade The seminal discovery of paclitaxel from endophytic fungus Taxomyces andreanae was a milestone in recognizing the immense potential of endophytic fungi as prolific producers of bioactive secondary metabolites of use in medicine, agriculture, and food industries. Following the discovery of paclitaxel, the research community has intensified efforts to harness endophytic fungi as putative producers of lead molecules with anticancer, anti-inflammatory, antimicrobial, antioxidant, cardio-protective, and immunomodulatory properties. Endophytic fungi have been a valuable source of bioactive compounds over the last three decades. Compounds such as taxol, podophyllotoxin, huperzine, camptothecin, and resveratrol have been effectively isolated and characterized after extraction from endophytic fungi. These findings have expanded the applications of endophytic fungi in medicine and related fields. In the present review, we systematically compile and analyze several important compounds derived from endophytic fungi, encompassing the period from 2011 to 2022. Our systematic approach focuses on elucidating the origins of endophytic fungi, exploring the structural diversity and biological activities exhibited by these compounds, and giving special emphasis to the pharmacological activities and mechanism of action of certain compounds. We highlight the tremendous potential of endophytic fungi as alternate sources of bioactive metabolites, with implications for combating major global diseases. This underscores the significant role that fungi can play in the discovery and development of novel therapeutic agents that address the challenges posed by prevalent diseases worldwide. Introduction Prominent amongst modern day healthcare challenges is the emergence of resistance among pathogenic microorganisms, novel occurrences of life-threatening viruses, and rising incidences of communicable and noncommunicable diseases. These medical challenges provide an urgent and compelling need to harness and leverage novel resources that offer sustainable solutions [1][2][3]. Natural products, which are metabolites or by-products derived from plants, animals, or microorganisms, have always been used to treat various kinds of 2 of 44 human ailments. The novel structures and frameworks of these compounds, combined with their broad-spectrum activities and potential as lead molecules, offer immense promise in therapeutic applications [4,5]. Natural products have the potential to be directly used as drugs or as building blocks for the synthesis of new drugs through combinatorial synthesis methods. They can either be utilized in their original form or modified to synthesize novel compounds with enhanced pharmacological properties. It is noteworthy that approximately 55% of the drugs that have been approved for clinical use in the last three decades can be traced back to natural products, whereas 58% of drugs have been developed by imitating natural product structures [6][7][8]. For centuries, plants have served as a key source of phytochemicals for drug discovery and development [5,9]. The serendipitous discovery of penicillin in 1928 by Sir Alexander Fleming from Penicillium chrysogenum marked the beginning of the golden era of antibiotics. The subsequent success of several lifesaving drugs obtained from microorganisms, such as the cholesterol biosynthesis inhibitor lovastatin from Aspergillus terrus and the immunosuppressant cyclosporine from Tolypocladium inflatum has brought about a significant change in drug discovery and development, shifting the focus from plants to microorganisms [1,2,10]. Since then, fungi have played a significant role in benefiting human welfare through the production of bioactive compounds that have been utilized as antimicrobial [3,4], anticancer [11,12], antioxidant [13], and immunomodulatory agents [14]. However, even after the pioneering discovery of penicillin ninety-five years ago, fungi continue to be the most underexplored biosource of natural products, particularly considering their vast biodiversity, unique biochemical properties, and significant biotechnological potential. This is despite the ongoing characterization of over a thousand fungal species annually, with several thousand more awaiting isolation and further characterization. As a result, the utility of fungal products largely remains unexplored and untapped, despite the impressive new taxonomic findings. In addition, the complexity of fungal biosynthetic pathways, as revealed by whole genome sequencing and subsequent genome mining of various fungal species, poses further challenges to harnessing the full potential of fungi [5,[15][16][17][18]. To overcome the limitations faced by existing methods in the field of fungal bioprospecting, it is essential to adopt novel screening strategies that can effectively identify fungi inhabiting distinct ecological environments. One potential strategy involves targeting fungi that establish mutualistic alliances with plants, residing within their living tissues without causing any apparent symptoms. This particular group of fungi, known as endophytes, holds great promise as a source of bioactive compounds [3,16,19,20]. Endophytic fungi secrete an array of bioactive compounds that serve multiple functions, such as stimulating plant growth, inducing defense mechanisms against pathogens, and serving as agents for remediating salt and drought stresses [3,[21][22][23][24]. This coevolution between endophytic fungi and their host plants results in the production of bioactive compounds which contribute in a variety of ways to plant-microbe interactions and can provide fitness benefits to the host plant ( Figure 1) [25][26][27][28]. Endophytic fungi establish their communication with their host plants through metabolic interactions [29,30]. According to the xenohormesis hypothesis [31], heterotrophic organisms such as fungi, under selective evolutionary pressure, develop the ability to sense stress-induced chemical cues from host plants and start producing analogous chemicals themselves. In essence they mimic the biological properties of the host plant [27,32]. In addition to synthesizing compounds that are analogous to host plant compounds, endophytic fungi also exhibit a vast repertoire of diverse secondary metabolites with intriguing biological and/or pharmaceutical properties. In the last thirty years, a wide range of bioactive compounds with potential in healthcare and medicine have been discovered from endophytic fungi. These compounds exhibit various properties such as antimicrobial [1,3,18], anticancer [33,34], antioxidant [35][36][37][38], anti-inflammatory [39,40], antidiabetic [41][42][43], and immunosuppressive activities [44][45][46][47]. The abundance of such biologically active metabolites derived from endophytic fungi highlights their importance as a valuable source of potential therapeutic substances [3][4][5]18,22,34] ties [44][45][46][47]. The abundance of such biologically active metabolites derived from en phytic fungi highlights their importance as a valuable source of potential therapeutic s stances [3][4][5]18,22,34] What Is an Endophyte? There exists a vast number of plant species on earth, exceeding 300,000, and each these plants hosts a diverse range of microorganisms, broadly categorized as either e phytes, endophytes, or pathogens [3,48]. Among them, endophytes comprise a dive group of ubiquitous, polyphyletic microorganisms that reside within plant cells or in intracellular space for at least a part of their life cycle, without showing any external m ifestation of their presence [22,[49][50][51][52]. Fossil records indicate that the microorganisms sociated with plants can be traced back over 400 million years to the Devonian peri suggesting that the alliance between plants and endophytes may have originated dur the early emergence of land plants on earth [27,30,53]. The term "endophyte" is deri from its literal meaning of "within the plants" ("endon" meaning within; "phyton" me ing plants) [54]. In 1866, the German botanist Anton de Bary coined the term "endophyt to describe organisms that live within plants without any visible symptoms; however, first endophytes were discovered in 1904 from a Eurasian darnel ryegrass, Lolium temu tum [23,55,56]. Endophytes can be found thriving in a range of ecological niches includ the Artic and the Antarctic regions, deserts, mangroves, rainforests, as well as marine a coastal ecosystems [9,24,51,57,58]. Endophytes exhibit diverse relationships with th host plants including symbiotic, benign commensal, decomposer, and latent pathoge interactions [3,59]. Once an endophyte successfully colonizes the internal tissue of a h plant, it enters a dormant phase that can persist for its entire lifecycle or for an extend duration until favorable conditions arise. During this period, endophytes remain inact or exhibit minimal metabolic activity, waiting for environmental cues that indicate availability of suitable conditions for growth and proliferation. This coevolutionary p cess creates a mutually beneficial relationship between the host plant and the endoph The host plant supplies vital nutrients and shelter to the endophyte that are required its survival, while the endophyte reciprocates by producing bioactive metabolites that hance the fitness of the host plant. The bioactive metabolites produced by endophytes p a crucial role in enabling host plants to withstand biotic and abiotic stresses, conserve w ter, and defend themselves against microbial, pest, and insect attack. As a result, en phytes play a vital role in plant symbiosis by providing protection to their host pla What Is an Endophyte? There exists a vast number of plant species on earth, exceeding 300,000, and each of these plants hosts a diverse range of microorganisms, broadly categorized as either epiphytes, endophytes, or pathogens [3,48]. Among them, endophytes comprise a diverse group of ubiquitous, polyphyletic microorganisms that reside within plant cells or in the intracellular space for at least a part of their life cycle, without showing any external manifestation of their presence [22,[49][50][51][52]. Fossil records indicate that the microorganisms associated with plants can be traced back over 400 million years to the Devonian period, suggesting that the alliance between plants and endophytes may have originated during the early emergence of land plants on earth [27,30,53]. The term "endophyte" is derived from its literal meaning of "within the plants" ("endon" meaning within; "phyton" meaning plants) [54]. In 1866, the German botanist Anton de Bary coined the term "endophytes" to describe organisms that live within plants without any visible symptoms; however, the first endophytes were discovered in 1904 from a Eurasian darnel ryegrass, Lolium temulentum [23,55,56]. Endophytes can be found thriving in a range of ecological niches including the Artic and the Antarctic regions, deserts, mangroves, rainforests, as well as marine and coastal ecosystems [9,24,51,57,58]. Endophytes exhibit diverse relationships with their host plants including symbiotic, benign commensal, decomposer, and latent pathogenic interactions [3,59]. Once an endophyte successfully colonizes the internal tissue of a host plant, it enters a dormant phase that can persist for its entire lifecycle or for an extended duration until favorable conditions arise. During this period, endophytes remain inactive or exhibit minimal metabolic activity, waiting for environmental cues that indicate the availability of suitable conditions for growth and proliferation. This coevolutionary process creates a mutually beneficial relationship between the host plant and the endophyte. The host plant supplies vital nutrients and shelter to the endophyte that are required for its survival, while the endophyte reciprocates by producing bioactive metabolites that enhance the fitness of the host plant. The bioactive metabolites produced by endophytes play a crucial role in enabling host plants to withstand biotic and abiotic stresses, conserve water, and defend themselves against microbial, pest, and insect attack. As a result, endophytes play a vital role in plant symbiosis by providing protection to their host plants against pathogenic threats and challenging environmental conditions [9,24,25,32,60]. Endophytes continuously adapt and evolve in response to biotic and abiotic stresses forming intricate interactions (bi-, tri-, or multipartite) with their host plant. This symbiotic relationship leads to the production of valuable natural products with therapeutic potential. These bioactive compounds, produced through the ongoing process of the strain development of endophytes, can be utilized directly or indirectly as therapeutic agents. The dynamic interplay between endophytes and their host plants gives rise to a diverse range of bioactive metabolites that holds promise for various therapeutic applications [17,25,61]. Through genetic recombination with the host plant, endophytes also acquire the ability to emulate the biological properties of their host plant and produce analogous bioactive metabolites. This proficiency in metabolism makes them a highly valuable resource for the exploration and discovery of natural bioactive metabolites [5,17,25]. Exploring Bioactive Metabolites from Endophytic Fungi: Unveiling Nature's Treasure Trove Endophytes, despite being isolated as early as 1904, remained largely overlooked for a considerable period of time. Apart from sporadic research, the biochemical research community did not pay much attention to endophytes until the discovery of Taxomyces andreanae, an endophytic fungus isolated from Pacific yew (Taxus brevifolia) in 1993. This fungus demonstrated an extraordinary capability to independently produce the highly successful anticancer drug taxol in its culture broth, resembling its host [62,63]. This breakthrough discovery initiated a global quest among researchers to delve deeper into the exploration of endophytic fungi with the aim of uncovering potential bioactive compounds [20,25,30,32,51,57,58,62,64,65]. Following this, significant findings emerged which unveiled the potential of endophytes to synthesize analogous bioactive metabolites with notable therapeutic properties. Compounds such as taxol, resveratrol, huperzine, camptothecin, podophyllotoxin, and vinca alkaloids were among those discovered from endophytic fungi, showcasing their ability to produce bioactive compounds that can be used as therapeutic agents for the treatment of diverse ailments, either through direct application or indirect utilization (Table 1, Figure 2) [23,25,52,[66][67][68][69]. Furthermore, endophytic fungi are also capable of producing a wide range of nonanalogous compounds that exhibit significant bioactivities. The bioactive metabolites derived from endophytes predominantly belong to the chemical class of alkaloids, cytochalasins, flavonoids, polyketides, steroids, and terpenoids [70,71]. The metabolites produced by endophytes have been found to display a wide range of pharmacological properties primarily encompassing antimicrobial, antineoplastic, anticancer, antioxidant, anti-inflammatory, antidiabetic, and antidepressant activities [2][3][4]22,25,34,52,72]. In addition, endophytes have been identified as a viable source of numerous enzymes such as amylase, catalase, laccase, lipase, and proteases that have significant clinical and industrial applications [40,58,73]. Thus, endophytic microorganisms represent a valuable reservoir of bioactive secondary metabolites with tremendous potential in the agrochemical and pharmaceutical industries [2,8,9,51,74,75]. This review highlights significant bioactive molecules discovered from endophytic fungi over the last decade, along with their potential applications in the treatment of various life-threatening diseases. The article presents a comprehensive analysis of 296 newly discovered compounds derived from endophytic fungi, characterized by novel or rare structures or skeletal frameworks across 290 journal articles published between 2011 and 2022. Furthermore, the article provides a concise overview of the origin of these endophytic fungi, the chemical structures of the compounds, and their corresponding biological activities. Anticancer Compounds from Endophytic Fungi Cancer is a complex and diverse group of diseases characterized by uncontrolled growth and spread of abnormal cells in the body. It is a major contributor to global mortality and the future outlook indicates an upward trend in cancer-related deaths. The Anticancer Compounds from Endophytic Fungi Cancer is a complex and diverse group of diseases characterized by uncontrolled growth and spread of abnormal cells in the body. It is a major contributor to global mortality and the future outlook indicates an upward trend in cancer-related deaths. The World Health Organization (WHO) has estimated that around ten million people died from cancer worldwide in 2020, and this figure is anticipated to reach 13.1 million by the year 2030 [95]. The most prevalent types of cancer globally include breast, cervical, colon, prostrate, oral, rectal, skin, and stomach cancer. Cancer incidences exhibit significant variation across countries and regions with higher rates observed in more developed nations. The incidence of cancer is influenced by several factors, including genetics, lifestyle choices, and access to healthcare [96]. The primary treatment options for the aforementioned types of cancer include surgery, radiation, immuno-and chemotherapy. While these treatments can be effective to a certain extent, they often come with significant side effects such as weakness, hair loss, cognitive issues, and increased vulnerability to infections. Additionally, some cancer cells develop resistance to these drugs, which diminishes the effectiveness of these therapies [97]. To address these challenges, ongoing research is dedicated to developing novel anticancer compounds and therapies that offer a precise targeting of cancer cells and have fewer side effects [98]. In recent years, there has been a considerable interest in fungal endophytes as a potential source of new drugs. This interest stems from the remarkable discovery of the anticancer drug "taxol" in the endophytic fungus T. andreanae isolated from the Pacific yew tree [62,99]. This breakthrough instigated a widespread initiative to systematically screen diverse plant species for the presence of taxol-producing endophytes. This approach has been successful in finding taxol-producing endophytes not only in the taxus plant but also in various other plant species. Extensive studies have identified taxol or its analogue-producing endophyte in various fungal genera, including Alternaria, Bartalinia, Fusarium, Lasiodiplodia, Metarhizium, Monochaetia, Pestalotiopsis, Penicillium, Phoma, Pithomyces, Seimatoantlerium, Sporormia, Trichothecium, Tubercularia, and Truncatella. Originally, taxol was found to be active against L-1210, P-388, and P-1534 leukemias whereas now, taxol is primarily used in combination with other anticancer drugs for the treatment of breast, ovarian, lung, and advanced testicular cancers [23,100]. Vinblastine and vincristine (also known as vinca alkaloids) are plant-based chemotherapeutic agents that exhibit therapeutic activity by binding to microtubule and spindle proteins, leading to cell-cycle arrest and apoptotic cell death in cancer cells. Initially isolated from Madagascar periwinkle plant (Catharanthus roseus), these vinca alkaloids have been widely employed in the treatment of various cancer types. The discovery of vinblastine and vincristine sparked a global quest to explore alternative sources of these valuable compounds [101,102]. Fungal endophytes such as Eutypella sp., Fusarium oxysporum, Nigrospora sphaerica, and Talaromyces radicus isolated from Madagascar periwinkle plant have been discovered to produce vinblastine and vincristine. These compounds have exhibited cytotoxic activity in a dose-dependent manner against HeLa, MCF-7, A-549, U-251, A-431, and MDA-MB 231 cancer cell lines [79,80,82,83]. Camptothecin, a pentacyclic quinolone alkaloid, is primarily sourced from the wood of Camptotheca acuminate (a Chinese ornamental plant) and the roots of Nothapodytes foetida [67,87,103]. Camptothecin is the third largest plant-based antineoplastic agent that executes its cytotoxic property by selectively inhibiting topoisomerase I, an enzyme which plays a vital role in DNA replication [23,102]. In recent years, several fungal endophytes such as Aspergillus sp. LY341, LY355, Alternaria burnsi, F. solani S-019, and Trichoderma atroviride LY357 have been found to produce camptothecin, which has a cytotoxic effect on human breast, lung, and ovarian cancer cell lines [86,87,104]. Podophyllotoxin is a highly valued aryltetralin lignin that serves as a precursor for the synthesis of anticancer drugs including etoposide, teniposide, and etopophos phosphate, which are clinically used for treating bronchial and testicular cancers [23]. Podophyllotoxin has a potent inhibitory effect on microtubule assembly, while its derivatives, etoposide and teniposide inhibit the activity of the topoisomerase enzyme II, resulting in cell-cycle arrest in the S phase [89]. Notably, certain endophytic fungi including Mucor fragilis TW5 and A. tenuissima have been identified as producers of podophyllotoxin, which exhibits cytotoxic activity against human colon, lung, and prostate cancer cell lines [89,105]. Xylaria psidii, an endophytic fungus isolated from the leaves of Aegle marmelos, yielded two notable compounds, xylarione A and (−) 5-methylmellein, which exhibited cytotoxic activity against MCF-7, MIA-Pa-Ca-2, NCI-H226, HepG2, and DU-145 cancer cell lines with an IC 50 value ranging from 16 to 37 µM [106]. Similarly, cytochalasin Q, a bioactive compound isolated from endophytic Xylaria sp. ZJWCF255 displayed cytotoxic activity against SMMC-772, MCF-7, and MGC80-3 cancer cell lines [34]. Furthermore, endophytic Chaetomium globosum isolated from Ginko biloba produced chaetoglobosin A, which showed remarkable cytotoxicity against HCT-116 cell lines with an IC 50 values in the range of 3.15-8.44 µM [107]. The identification of endophytic fungi as a lucrative source of anticancer drugs has opened up new possibilities for drug development. These fungi have been found to possess a wide range of bioactive compounds that hold potential in the fight against cancer. However, it is crucial to emphasize here that the production and development of these compounds are still in the early stages of investigation. Further studies are needed to elucidate their precise mechanisms of action, evaluate their safety profiles, and assess their suitability for clinical use. Further research is necessary to unlock the full potential of endophytic fungi as a viable source of effective and safer anticancer drugs (Table 2; Figure 3). Antioxidant Compounds from Endophytic Fungi Free radicals are unstable molecules that are either produced naturally in the body as a byproduct of metabolism or can be formed by external factors such as UV light, pesticides, drugs, smoking, and alcohol. These free radicals can damage cells and lead to var- Antioxidant Compounds from Endophytic Fungi Free radicals are unstable molecules that are either produced naturally in the body as a byproduct of metabolism or can be formed by external factors such as UV light, pesticides, drugs, smoking, and alcohol. These free radicals can damage cells and lead to various diseases such as diabetes, Down's syndrome, degenerative disease, Alzheimer's disease, Parkinson's disease, and cardiovascular disorders [129][130][131]. Antioxidant compounds protect cellular damage by neutralizing these free radicals and preventing them from causing cellular damage. Antioxidant compounds also play a significant role in preventing cancer, as they can react with and neutralize the free radicals that contribute to the formation of cancer cells. Thus, it is important to develop new antioxidant compounds, as they can help to stabilize free radicals and prevent cellular damage, thus improving human health and preventing degenerative and other diseases [132]. Studies have shown that endophytic fungi can produce a wide variety of compounds with strong antioxidant activities, and some of them have been isolated and characterized, such as phenolic acids, xanthones, flavonoids, terpenoids, and polyketides [133]. The bioprospection of endophytic fungi is a promising area, and new antioxidant compounds from fungal endophytes are continuously being discovered and characterized. Anofinic acid obtained from endophytic A. tubenginses ASH4 showed potential antioxidant and anticancer activities [134]. Endophytic Aspergillus sp. MFLUCC16-0603, MFLUCC16-0614, and Nigrospora sp. MFLUCC16-0605 isolated from Ocimum basilicum exhibited antioxidant activity with IC 50 values ranging between 11.75 and 17.39 mg/mL, respectively [135]. Similarly, endophytic A. alternata and P. citrinum isolated from Azadirachta indica have been found to have potential antioxidant activity, with IC 50 values of 38-52.13 µg/mL, respectively [136]. Other endophytic species, such as Chaetomium sp., Colletotrichum sp., Curvularia sp., and Trichoderma sp., isolated from similar host plants, have also exhibited antioxidant activity ranging from 31 to 69% [137]. Five endophytic fungal isolates PAL 01-B2, PAL 01-D2, PAL 04-R2, PAL 11-B1, and PAL 14-D3 possessed strong antioxidant activities with IC 50 value ranging between 5.26 and 14.06 µg/mL, respectively [138]. Potential antioxidant properties were also demonstrated by endophytic Aspergillus sp., Alternaria sp. (ML4), Chaetomium sp., Penicillium sp., and Phomopsis sp. GJJM07 isolated from Calotropis procera, Eugenia jambolana, Mesua ferrea, Trigonella foenum-graecum, and Triticum durum, respectively [139][140][141][142]. Fermentation extracts of fungal endophytes ZA 163, MO 211, LO 261, FE 082, and FE 084 associated with Nigerian ethnomedicinal plants Albizia zygia, Millettia thonningii, Alchornea cordifolia, and Ficus exasperate were found to produce pyrogallol, dl-alpha-tocopherol, Alpha tocospiro, linoleic acid, 9-octadecenamide, lupeol, and 9-octadecenoic acid (Z), which exhibited antioxidant activity [143]. Similarly, the fungal extracts of Fusarium SaR-2 and Alternaria SaF-2 have significant antioxidant properties with 90.14% and 83.25% free-radical scavenging activity, respectively [144]. Furthermore, extracts of Chaetomium globosum associated with Adiantum capillus showed 99% free-radical scavenging activity at a concentration of 100 µg/mL [145] (Table 3) (Figure 4). Anti-Inflammatory Compounds from Endophytic Fungi Inflammation is a multifaceted aspect of the immune response that arises in response to various factors such as pathogens, cellular injury, toxins, and radiation. It can manifest as either a short-term immediate reaction or as a long-lasting persistent condition. Inflammation has the potential to impact a wide range of body organs including the heart, pancreas, liver, kidney, lungs, brain, intestinal tract, and reproductive system. The underlying cause of inflammation can either be infectious or noninfectious and if left unresolved, it can result in tissue damage or contribute to disease development, depending upon the causative agent involved [154,155]. Studies have shown that the metabolome of endophytic fungi includes anti-inflammatory compounds similar to their host and are thus believed to be a potential source of agents for combating inflammation and improving human health. Lasiodiplactone A, derived from the marine mangrove plant Acanthus ilicifolius and produced by the endophytic fungus Lasiodiplodia theobromae demonstrated significant anti-inflammatory activity by inhibiting the production of nitric oxide (NO) in RAW 264.7 cells stimulated with lipopolysaccharide with an IC 50 of 23.5 µM. In addition, it also exhibited inhibitory activity against α-glucosidase with an IC 50 value of 29.4 µM [154]. Similarly, Botryosphaerin B, derived from the endophytic fungus Botryosphaeria sp. SCSIO KcF6 in the mangrove plant Kandelia candel showed an inhibitory effect on cyclooxygenase (COX)-2 activity with a significant IC 50 value of 1.12 mM [155]. Cyclonerodial B obtained from the endophytic fungus Trichoderma sp. Xy24 isolated from Xylocarpus granatum exhibited anti-inflammatory properties by suppressing the production of nitric oxide (NO) in BV2 microglia cells. This compound also has potential therapeutic applications in the treatment of neurodegenerative diseases such as Parkinson's and Alzheimer's [156]. Additionally, pretreated extract derived from the fungal endophytes Cytospora rhizophorae isolates HAB10R12, HAB16R13, HAB16R14, HAB16R18, and HAB8R14 obtained from Cinnamomum porrectum had inhibitory effects on the production of NO, interleukin (IL)-6, and TNF-α by activated BV2 microglia cells [157]. Furthermore, various endophytic fungi such as Aspergillus niger, Rhizopus oryzae, Dendryphion nanum, Pleospora tarda, and Penicillium sp. also showed anti-inflammatory properties. These fungi have demonstrated the ability to inhibit the activity of COX 1, COX 2, and 5-lipoxygenase which are involved in the inflammatory process. Additionally, they also produce herbarin, known for its anti-inflammatory activity. Studies also suggested that the anti-inflammatory activity of these fungi are dose-dependent, and they have been found to inhibit protein and albumin denaturation [158][159][160] (Table 4, Figure 5). Anti-Inflammatory Compounds from Endophytic Fungi Inflammation is a multifaceted aspect of the immune response that arises in response to various factors such as pathogens, cellular injury, toxins, and radiation. It can manifest as either a short-term immediate reaction or as a long-lasting persistent condition. Inflam- Dendryphion nanum Ficus religiosa Herbarin Cytokines TNF-α and IL-6 0.60 µM [158] NO: nitric oxide, COX: cyclooxygenase, TNF: tumor necrosis factor. Antidiabetic Compounds from Endophytic Fungi Diabetes mellitus (DM) is a chronic metabolic disorder marked by elevated levels of glucose in the blood (hyperglycemia) and the disruption of carbohydrate, protein, and fat metabolism. DM is linked to various complications such as cardiovascular disorders, retinopathy, nephropathy, and neuropathy. The prevalence of DM is rising, and projections indicate that by 2030 about 522 million individuals will be affected worldwide. India in particular is expected to experience a high burden of DM cases in the future. One management strategy for DM is the inhibition of α-glucosidase and α-amylase enzymes. These enzymes play a crucial role during the breakdown of carbohydrates during digestion. By slowing down their activity, the rate of carbohydrate digestion and subsequent absorption of glucose into the blood stream can be reduced, leading to better control of blood glucose levels. Inhibiting these enzymes has proven to be an effective approach in managing DM and mitigating hyperglycemia. Acarbose and miglitol are examples of drugs that specifically inhibit α-glucosidase activity, thus helping to regulate blood levels in individuals with DM. Recent studies have indicated that fungal endophytes have the potential to serve as valuable source of inhibitors for α-glucosidase and α-amylase. The compounds S (+)-2 Antidiabetic Compounds from Endophytic Fungi Diabetes mellitus (DM) is a chronic metabolic disorder marked by elevated levels of glucose in the blood (hyperglycemia) and the disruption of carbohydrate, protein, and fat metabolism. DM is linked to various complications such as cardiovascular disorders, retinopathy, nephropathy, and neuropathy. The prevalence of DM is rising, and projections indicate that by 2030 about 522 million individuals will be affected worldwide. India in particular is expected to experience a high burden of DM cases in the future. One management strategy for DM is the inhibition of α-glucosidase and α-amylase enzymes. These enzymes play a crucial role during the breakdown of carbohydrates during digestion. By slowing down their activity, the rate of carbohydrate digestion and subsequent absorption of glucose into the blood stream can be reduced, leading to better control of blood glucose levels. Inhibiting these enzymes has proven to be an effective approach in managing DM and mitigating hyperglycemia. Acarbose and miglitol are examples of drugs that specifically inhibit α-glucosidase activity, thus helping to regulate blood levels in individuals with DM. Recent studies have indicated that fungal endophytes have the potential to serve as valuable source of inhibitors for α-glucosidase and α-amylase. The compounds S (+)-2 cis 4-trans abscisic acid and 7 hydroxyl abscisic acid, 4 deshydroxyl, and altersolanol A isolated from endophytic Nigrospora oryzae associated with Combretum dolichopetalum demonstrated a significant reduction in blood sugar levels in mice with induced diabetes. S (+)-2 cis 4-trans abscisic acid specifically showed antidiabetic properties by enhancing the activity of peroxisome proliferator-activated gamma receptor (PPAR γ) in immune cells [173]. Thielavins A, J, and K obtained from endophytic fungal isolate MEXU 27095 exhibited a dose-dependent inhibition of α-glucosidase, with IC 50 values of 15.8, 22.1, and 23.8 µM, respectively [174]. Likewise, Aspergiamides A and F, isolated from Aspergillus sp. derived from Sonneratia apetala, demonstrated α-glucosidase inhibitory activity with IC 50 values of 40 and 83 µM, respectively [175]. Peptides produced by Aspergillus awamori significantly inhibited the activity of both α-glucosidase and α-amylase with IC 50 values of 3.75 and 5.62 µg/mL, respectively. These inhibitors were stable over a wide range of pH and temperature conditions and exhibited nonmutagenic properties [176]. Fungal endophytes derived from medicinal diabetic plants in Uzbekistan exhibited a remarkable 60-82% inhibitory activity against α-amylase. Recently, K-10, a polymethoxylated flavone methanolic extract from endophytic Aspergillus egypticus-HT166S isolated from Helianthus tuberosus showed an inhibition of α-amylase similar to a reference standard (acarbose) in lab conditions [177,178]. Similarly, endophytic isolates from Stemphylium globuliferum PTFL005 and PTFL011 exhibited inhibitory activity against α-glucosidase with IC 50 values of 17.37 and 10.71 µg/mL, respectively. Additionally, Stemphylium globuliferum PTFL005 and PTFL006 demonstrated encouraging α-amylase inhibitory activity with IC 50 values of 15.48 and 13.48 µg/mL, respectively [179]. Endophytic Alternaria destruens isolated from Calotropis gigantea exhibited a weak inhibition of α-amylase (31%) and a strong inhibition of α-glucosidase (93%). Similarly, endophytic Xylariaceae sp. QGS01, Penicillium citrinum, and Colletotrichum sp. were also reported as potential inhibitors of α-glucosidase, suggesting their possible use in the management of DM [41,42,180]. Antidiabetic properties have been observed in several marine-and mangrove-derived fungi. Studies have identified certain compounds such as eremophilane sesquiterpenes from endophytic Xylaria sp., and isopimarane diterpene and 11-deoxydiapothein A from Epicoccum sp. HS-1 significantly inhibited α-glucosidase enzyme [181,182]. Similarly, tripalmitin, a mixed inhibitor derived from mangrove endophytic Zasmidium sp. strain EM5-10 exhibited significant inhibitory activity against α-glucosidase compared to acarbose. In silico studies of tripalmitin predicted that it bound to the same site as acarbose as well as an additional allosteric site in human intestinal α-glucosidase [183]. The aforementioned studies indicate that endophytes hold promise as novel inhibitors of α-amylase and αglucosidase, which can contribute to the improved management of DM. By harnessing these endophytes, it may be possible to develop effective strategies for better control and treatment of DM (Table 5, Figure 6). Immunosuppressive Compounds from Endophytic Fungi Immunosuppressive medications are essential in preventing, suppressing, or minimizing organ rejection in transplant patients. As a result, they are of utmost importance in effectively managing autoimmune diseases such as lupus, psoriasis, insulin-dependent diabetes, and rheumatoid arthritis. Despite their effectiveness, these medications are as- Immunosuppressive Compounds from Endophytic Fungi Immunosuppressive medications are essential in preventing, suppressing, or minimizing organ rejection in transplant patients. As a result, they are of utmost importance in effectively managing autoimmune diseases such as lupus, psoriasis, insulin-dependent diabetes, and rheumatoid arthritis. Despite their effectiveness, these medications are associated with potential side effects, emphasizing the necessity to seek safer alternatives that can offer effective immune modulation while minimizing adverse effects [185,186]. Fungal endophytes present a promising and innovative alternative source of immunosuppressive agents and have the potential to be developed into new therapeutic drugs [187]. Recent studies have found that certain compounds of endophytic origin, such as colutellin A, dibenzofurane, lipopeptide, sydoxanthone A and B, subglutinol A and B, and 13-Oacetylsydowinin B have potent immunosuppressive properties. These findings open new possibilities for the development of novel immunosuppressive drugs. However, it is important to note that these drugs are in the early stages of investigation, and further studies are warranted to assess their safety, effectiveness, and potential side effects [5]. Two endophytic fungi (PGS1 and NLL3) isolated from Psidium guajava and Newbouldia laevis, respectively, produced citrinin, nidulalin, p-hydroxybenzoic acid, and cyclopenin. These compounds have been associated with immunosuppressant properties [188]. Similarly, a chemical analysis of endophytic fungus Mycosphaerella nawae ZJLQ129 derived from Smilax china leaves demonstrated the presence of a novel amide derivative (−)mycousnine enamine. This derivative was found to selectively inhibit T-cell proliferation by blocking the expression of surface activation antigens CD25 and CD69. These findings indicate that endophytic fungi have the potential to serve as a valuable source of immunosuppressants that exhibit a high efficacy and low toxicity [189]. Similarly, the endophytic fungus Penicillium sp. ZJ-SY2, which was found in association with the mangrove species Sonneratia apetala, produces a collection of nine polyketides that include two novel benzophenone derivatives named peniphenone and methyl peniphenone, as well as seven xanthones. These compounds demonstrated potent immunosuppressive properties, with IC 50 values ranging from 5.9 to 9.3 µg/mL [190]. Endophytic Fusarium subglutinans, isolated from Tripterygium wilfordii, yielded subglutinol A and B, which have been reported to possess immunosuppressive properties [191]. Likewise, endophytic fungus Albifmbria viridis isolated from Chinese medicinal plant produced Albifpyrrols B, specifically inhibited the proliferation of B-lymphocyte cells induced by lipopolysaccharides (LPS) with an IC 50 value of 16.16 µM [47]. The endophytic Phomopsis sp. S12 derived compound libertellenone J has also been found to have notable immunosuppressive properties. It effectively reduces the production of NO, IL-1β, IL-6, and TNF-α with IC 50 values ranging from 2.2 to 10.2 µM. In addition, it also decreases the expression of iNOS, IL-1β, IL-6, and TNF-α mRNA in LPSactivated macrophages, with IC 50 values ranging from 3.2 to 15.2 µM [192]. Furthermore, a fermentation extract of endophytic Botryosphaeria dothidea BAK-1 isolated from Kigelia africana demonstrated a dose-dependent suppression of T-cell proliferation by 50% and TNF-α production by 55% [193]. These significant reports inspire the further exploration of fungal endophytes for new immunosuppressive agents [192] (Table 6, Figure 7). Pestalotiopsis sp. HHL-14 Antimicrobial Compounds from Endophytic Fungi The emergence of drug resistance among disease causing microorganisms is a burgeoning issue that needs urgent action. Infectious diseases are among the leading causes of deaths after cardiovascular disorders and cancers, as they account for 13.7 million deaths globally (13.6% of total global deaths) (Institute of Health Metrics and Evaluation 2019). The COVID-19 outbreak is a prime example of this situation, caused by the spread of a novel coronavirus. This virus has infected over 600 million individuals and tragically caused the death of more than 6.5 million people across the globe [1]. To address this pressing issue, there is a continuous quest to discover novel antimicrobial agents that are both effective and have reduced or minimal side effects. Endophytic fungi have been well recognized for their ability to produce a diverse array of secondary metabolites such as alkaloids, terpenoids, flavonoids, and polyketides. These compounds have demonstrated antimicrobial activity against various pathogenic microorganisms such as Staphylococcus aureus, Pseudomonas aeruginosa, Klebsiella pneumonia, Shigella flexneri, Enterococcus faecalis, Escherichia coli, Salmonella typhi, Bacillus subtilis, Saccharomyces cerevisiae, Candida albicans, F. oxysporum, human immunodeficiency virus (HIV), herpes simplex virus (HSV) and influenza virus (H1N1) [198][199][200][201][202][203]. In recent years, numerous bioactive metabolites have been isolated from endophytic fungi, exhibiting profound antimicrobial activities. Table 7 provides a comprehensive overview of these antimicrobial agents, highlighting their antibacterial, antifungal, and antiviral properties. Fumigaclavine C and fraxetin produced by A. fumigatus obtained from Ceriops decandra exhibited strong antibacterial activity against E. coli, Micrococcus luteus, S. aureus, and P. aeruginosa [204]. Antibacterial activity Endophytic Acremonium sp. MER V1 and Chaetomium sp. MER V7 isolated from Avicennia marina showed antiviral activity against hepatitis C virus. However, their fusant MER V6270 showed a stronger inhibition of hepatitis C virus as compared to individual fungus Acremonium sp. MER V1 and Chaetomium sp. MER V7 [257]. Phomanolide B obtained from endophytic Phoma sp. demonstrated antiviral properties towards influenza virus, whereas a novel bioactive compound Aspulvinone E, obtained from endophytic fungi A. terreus displayed strong antiviral activity against HIV [245,246]. Pestalotiopsis thea is an endophytic fungus that produces bioactive metabolites such as chloroisosulochrin, ficipyrone A and pestheic acid. Amongst them, chloroisosulochrin displayed maximum antirespiratory syncytial viral (RSV) inhibitor activity, whereas the other two compounds exhibited moderate activities against the virus [250]. Endophytic Pleospora tarda secreted alternariol and alternariol-(9)-methyl ester that showed moderate inhibitory activity against HSV (40%). Furthermore, fungal endophytes such as Nigrospora sphaerica, Acremonium strictum, Phoma leveillei, Aspergillus flavus, Chaetomium globosum, Mucor fuscus, Acremonium strictum, and Penicillium chrysogenoum, which were isolated from Chiliadenus montanus, Launea spinosa, Euphorbia sancta, Stachys aegyptiaca, Hypericum sinaicum, Stachys aegyptiaca, and Launea spinose, respectively, displayed weak to moderate (2-14%) activity against HSV [145]. COVID-19 is a new viral pandemic disease that originated in China and has spread to all countries worldwide. Currently, there is no specific drug available to treat COVID-19, and management is mainly focused on supportive care such as vitamin supplements, antibiotics, and oxygen therapy. Some researchers have proposed the possibility that endophytes may possess antiviral properties that could be effective against novel coronaviruses. In a study by [208], it was found that crude ethyl acetate extract derived from endophytic Curvularia papendorfii demonstrated potent antiviral activity against human coronavirus HCoV229E and feline coronavirus FCV F9. Furthermore, in another study, it was observed that fungal endophytes produced Aspergillide B1 and 3a-Hydroxy-3,5-dihydromonacolin L compounds. These compounds exhibited the highest binding energy scores when interacting with the protease (Mpro) of the novel coronavirus, indicating their potential as inhibitors against the virus [245]. However, it is crucial to emphasize that these findings are derived from preclinical studies and additional research is necessary to validate their effectiveness in in vivo settings and establish optimal dosage and administration protocols. Additionally, conducting clinical trials would be necessary to assess the safety and efficacy of fungal endophytes as a potential bioresource in the treatment of COVID-19 (Table 7, Figure 10). dida tropicalis, Curvularia, and Fusarium species [237]. Similarly, endophytic Lophodermium nitens DAOM 250027 isolated from Pinus strobus produces (7R)-(-)-methoxysydonol and its derivatives (7R,7′R)-(-)-pyrenophorin, which showed antifungal activity against S. cerevisiae [199]. Furthermore, Phialomustin C and D isolated from endophytic Phialophora mustea in Crocus sativus exhibited antifungal activity against C. albicans with an IC50 values of 14.3 and 73.6 µM, respectively [241] (Table 7, Figure 9). Antiprotozoal Compounds from Endophytic Fungi Protozoan parasites such as Tryanosoma cruzi, Plasmodium berghei, Plasmodium falciparum, and Leishmania amazonensis cause a range of diseases including Chagas disease, malaria and leishmaniasis. These diseases are vector-borne and are transmitted to humans through the bite of infected mosquitoes or flies [258]. They are classified as neglected diseases by the WHO, and primarily affect low-income areas, receiving limited attention in terms of research and development. In addition, the current drugs available for treating these diseases have significant limitations such as poor effectiveness, toxicity, drug resistance, and high cost. As a result, there is an urgent need to find new drugs that are effective, safer, and affordable. To address these issues, efforts are being made to explore different strategies such as repurposing existing drugs, screening chemical libraries, and developing new candidates through targeted or natural product-based approaches [259]. Studies suggest that the endophytic fungi derived from medicinal plants such as Artemisia annua, Cinchona calisaya, and Markhamia platycalyx have been found to produce bioactive compounds with inhibitory properties against the above-mentioned parasites. Notably, endophytic Nigrospora oryzae Cf-298113, isolated from the roots of Triticum sp., secrete pipecolisporin, which has potent inhibitory activity against P. falciparum (3.21 µM) and T. cruzi (8.68 µM) [260]. The antiplasmodial activity of endophytic Aspergillus terrus, Penicillum commune, P. chrysogenum, and Talaromyces piophilus isolated from A. annua has been investigated. Among these, the fermentation extract of P. commune and P. chrysogenum inhibited P. falciparum with IC50 values of 1.1 and 3.3 µg/mL, respectively. The extract from Talaromyces strains showed a moderate activity with IC50 values of 7.6-9.9 µg/mL, whereas the extract from A. terreus displayed a lower activity with an IC50 of 35 µg/mL [261]. In addition, two endophytic fungal strains (IP-2 and IP-6) isolated from A. annua demonstrated antiplasmodial activity with IC50 values of 30 and 42 µg/mL, respectively, whereas 19,20 epoxycytochalasin C derived from the ethyl acetate extract of endophytic Nemania Antiprotozoal Compounds from Endophytic Fungi Protozoan parasites such as Tryanosoma cruzi, Plasmodium berghei, Plasmodium falciparum, and Leishmania amazonensis cause a range of diseases including Chagas disease, malaria and leishmaniasis. These diseases are vector-borne and are transmitted to humans through the bite of infected mosquitoes or flies [258]. They are classified as neglected diseases by the WHO, and primarily affect low-income areas, receiving limited attention in terms of research and development. In addition, the current drugs available for treating these diseases have significant limitations such as poor effectiveness, toxicity, drug resistance, and high cost. As a result, there is an urgent need to find new drugs that are effective, safer, and affordable. To address these issues, efforts are being made to explore different strategies such as repurposing existing drugs, screening chemical libraries, and developing new candidates through targeted or natural product-based approaches [259]. Studies suggest that the endophytic fungi derived from medicinal plants such as Artemisia annua, Cinchona calisaya, and Markhamia platycalyx have been found to produce bioactive compounds with inhibitory properties against the above-mentioned parasites. Notably, endophytic Nigrospora oryzae Cf-298113, isolated from the roots of Triticum sp., secrete pipecolisporin, which has potent inhibitory activity against P. falciparum (3.21 µM) and T. cruzi (8.68 µM) [260]. The antiplasmodial activity of endophytic Aspergillus terrus, Penicillum commune, P. chrysogenum, and Talaromyces piophilus isolated from A. annua has been investigated. Among these, the fermentation extract of P. commune and P. chrysogenum inhibited P. falciparum with IC 50 values of 1.1 and 3.3 µg/mL, respectively. The extract from Talaromyces strains showed a moderate activity with IC 50 values of 7.6-9.9 µg/mL, whereas the extract from A. terreus displayed a lower activity with an IC 50 of 35 µg/mL [261]. In addition, two endophytic fungal strains (IP-2 and IP-6) isolated from A. annua demonstrated antiplasmodial activity with IC 50 values of 30 and 42 µg/mL, respectively, whereas 19,20 epoxycytochalasin C derived from the ethyl acetate extract of endophytic Nema-nia sp. UM10M showed a relatively weak antiplasmodial activity [262,263]. In addition, endophytic P. citrinum AMrb11 and Neocosmospora rubicola AMb22 exhibited potent antiplasmodial activity against both chloroquine-sensitivePf3D7 and chloroquine-resistant PfINDO/PfDd2 strains of P. falciparum, with IC 50 values ranging from 0.39 to 1.92 µg/mL for Neocosmospora rubicola AMb22 and 0.84-0.93 µg/mL for P. citrinum Amrb11 [264]. Moreover, a fermentation extract of endophytic Aspergillus flocculus yielded 3-hydroxymellein and dorcinol, which demonstrated significant inhibitory effects of 56 and 97% against the sleeping-sickness-causing parasite T. cruzi. The antitrypanosomal activity of A. flocculus is believed to be attributed to the synergistic effects of active steroidal compounds such as campesterol, ergosterol, and ergosterol peroxide [265]. Similarly, lead extracts obtained from endophytic isolates sourced from Antarctic angiosperms, particularly Deschampsia antartica, were tested for their ability to inhibit the proliferation of L. amazonensis. The IC 50 values of these extracts ranged from 0.2 to 125 µg/mL. Notably, Alternaria, Cadophora, Herpotrichia, and Phaeosphaeria spp. exhibited over 90% killing of L. amazonensis [266]. Recently, an in silico approach was employed to investigate the antileishmanial activity of epicoccamide derivatives A-D, which are of endophytic origin. These derivatives interacted with the active site of the enzyme through hydrogen bonds and hydrophobic interactions, leading to their stabilization. Epicoccamide derivatives exhibited high bonding energies with the trypanothione reductase of −13.31, −13.44, −13.31, and −13.32 kcal/mol, respectively [267,268] (Table 8, Figure 11). Prospects and Challenges Over the last few years, endophytic fungi have attracted significant attention in natural-product-based drug discovery due to their inherent capability to produce secondary metabolites as a source of novel drugs with low toxicity for treating various human ailments [33,34]. However, despite the progress made in studying endophytic fungi, only a fraction of endophytes have been explored so far (about 1%), and the vast majority of these organisms remain untapped and uncharacterized, with great potential for discovering new bioactive compounds [5,9,23,37,52]. To effectively isolate endophytes with significant bioactivity, a selection of host plant and its ecological niche is crucial. Plants that inhabit areas with high biodiversity, particularly those with endemic plant species, are more E. nigrum -Epicoccamide Leishmania sp. Prospects and Challenges Over the last few years, endophytic fungi have attracted significant attention in natural-product-based drug discovery due to their inherent capability to produce secondary metabolites as a source of novel drugs with low toxicity for treating various human ailments [33,34]. However, despite the progress made in studying endophytic fungi, only a fraction of endophytes have been explored so far (about 1%), and the vast majority of these organisms remain untapped and uncharacterized, with great potential for discovering new bioactive compounds [5,9,23,37,52]. To effectively isolate endophytes with significant bioactivity, a selection of host plant and its ecological niche is crucial. Plants that inhabit areas with high biodiversity, particularly those with endemic plant species, are more likely to harbor endophytes with novel chemical entities. When selecting a host plant for endophyte isolation, preference should be given to plants with known medicinal properties. This approach enhances the likelihood of identifying endophytes that produce bioactive compounds relevant to human health [4,34,48,51]. Furthermore, establishing connections between fungal metabolites and plant genomics enhances our understanding of the biosynthetic pathways involved in the process, justifying the production of the metabolites based on scientific knowledge and evidence, rather than relying upon unproven hypotheses [1,5,18,22] The production of bioactive compounds from fungal endophytes on an industrial scale is a complex and arduous task, necessitating advanced and efficient approaches [23]. Cutting-edge techniques such as CRISPR-Cas9 and epigenetic modifiers show promise in enhancing bioactive compound production. Moreover, several other strategies such as optimizing culture parameters, employing elicitors, and utilizing coculture fermentation have been successfully employed in laboratory conditions to augment the production of bioactive compounds from fungal endophytes [43]. However, isolating and characterizing promising fungal endophytes capable of producing bioactive compounds has always posed significant challenges. The integration of molecular approaches and bioinformatics, including phylogenetic studies, offers a potential solution by facilitating the precise delineation of fungal strains at the species level [23]. Under in situ conditions, endophytes coexist and interact with various other organisms, which significantly influences the production of secondary metabolites. However, when studied in in vitro conditions, endophytes are typically cultured under axenic conditions, devoid of these natural interactions. Therefore, it is essential to explore the interactions among endophytes, their host plants, and other associated microorganisms to fully harness their potential for the production of bioactive compounds [21,32,37]. These interactions are highly sensitive to culture conditions, offering an opportunity to optimize in vitro conditions and create an environment that stimulates the production of the desired bioactive compounds [4,9,34]. By adjusting culture conditions, media composition, aeration rate, and temperature, it is feasible to produce a specific desired compound. Furthermore, cocultivating endophytes in the presence of other microorganisms triggers the activation of biosynthetic pathways, leading to the synthesis of bioactive metabolites which are not produced when endophytes are cultured individually. Consequently, extensive research will be necessary to gain a comprehensive understanding of endophytes' biosynthetic capabilities. By developing suitable cocultivation methods and optimizing culture conditions, a consistent and efficient production of desired bioactive compounds may be possible from endophytic fungi in the future [18,21,43]. The process of discovering natural products traditionally involves bioprospecting various organisms and conducting laboratory screening programs, resulting in complex data. However, this approach often faces high attrition rates and challenges. To overcome these issues, artificial intelligence (AI) and machine language (ML) are increasingly being employed. The recent breakthroughs in AI, particularly in ML, have revolutionized the field of natural-product-based drug discovery programs. AI tools have demonstrated their effectiveness in uncovering hidden patterns, classifying objects, and clustering compounds based on their characteristics [280,281]. AI tools such as LeafNet, LeafSnap, ResNet26, IDBac, and SPeDE have been developed to assist in taxonomic identification, enabling the selection of novel organisms. For genome mining and chemical dereplication [282][283][284][285][286], AI tools such as ANtiSMASH, MIBiG, IMG-ABC, NRPro, CHEM, ELINA, and DEREP-NP have proven valuable. These tools help in the analysis and interpretation of genomic data, allowing researchers to identify potential gene clusters responsible for the biosynthesis of bioactive compounds. Furthermore, they aid the dereplication process by comparing chemical structures and identifying known compounds, thereby facilitating the selection of novel organisms with unique chemical profiles [287][288][289][290][291][292][293]. In the field of target identification, AI tools such as AutoDock, Schrodinger, SDiDER, and BANDIT play a crucial role. These tools utilize molecular docking and ligand-based approaches to predict the interactions between bioactive compounds and target proteins. By simulating the binding process, potential targets can be identified and the design of new compounds optimized [294][295][296][297]. The integration of AI tools into bioactive compound discovery has significantly enhanced the efficiency and accuracy of the process, accelerating the identification and development of bioactive compounds with therapeutic potential [280]. Conclusions The microbial world of plants holds great promise for future medicine. The scientific community has directed considerable attention towards fungal endophytes, recognizing their potential to synthesize bioactive compounds with a wide range of properties that may be antimicrobial, anticancer, antioxidant, anti-inflammatory, antidiabetic, immunomodulatory, and cardio-protective. This underscores that fungal endophytes are a bioresource for the development of novel drugs and other biotechnology products. Studies have shown that a significant portion (about 51%) of the bioactive metabolites sourced from endophytic fungi possess unique chemical structures. This emphasizes the existence of a vast and untapped reservoir, holding great potential for future exploration and development. The field of fungal endophytic biology has experienced significant technological advancement that has opened fresh avenues for the isolation and characterization of novel bioactive compounds. These advances encompass sophisticated molecular techniques to isolate and characterize endophytic fungi, as well as the development of novel methods to isolate bioactive compounds, both culture-dependent and culture-independent. These modern methods have greatly enhanced the efficacy and precision of isolation processes, enabling the discovery of previously unknown bioactive compounds from endophytic fungi. Moreover, the integration of bioinformatics tools and computational biology approaches has played a pivotal role in the discovery and characterization of bioactive compounds from endophytic fungi. These tools have provided valuable insights into the biosynthesis and regulation of secondary metabolites within endophytic fungi, facilitating the identification of new gene clusters and biosynthetic pathways associated with bioactive compound production. By leveraging these technological advances, researchers are now able to delve deeper into the untapped potential of endophytic fungi and uncover a wealth of promising bioactive compounds. However, despite the advancements in the field, the exploration of endophytic fungi for bioactive compounds is still in its early stages. Consequently, there is a pressing need to realign our research priorities towards biotechnological advances to expedite the screening and discovery of new biomolecules. However, conducting a thorough review of the literature and documentation regarding host plants, biosynthetic machineries, and their mechanism of action can yield valuable insights for potential explorations and bioprospecting endeavors. This comprehensive understanding offers opportunities to harness endophytic fungi as a sustainable and renewable source of bioactive compounds, contributing to human health and addressing the challenges of antibiotic resistance.	2023-07-30T06:17:10.055Z	2023-06-25T00:00:00.000	{ "year": 2023, "sha1": "5dab61b1767a3153ed96c9664f3019c3ec5fd4e6", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2218-273X/13/7/1038/pdf?version=1687829005", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "0f8f020d480387c43ecb9925a924bbfedc5c3380", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
265070043	pes2o/s2orc	v3-fos-license	Unveiling the potential effects of acetylsalicylic acid: insights into regeneration in endometrial stem cells Background Although acetylsalicylic acid has been widely used for decades to treat and prevent various diseases, its potential effects on endometrial receptivity and subsequent pregnancy rates are still controversial due to conflicting data: many reports have shown positive effects of acetylsalicylic acid, whereas others have found that it has no effect. Furthermore, the direct effects of acetylsalicylic acid on various functions of normal endometrial cells, especially endometrial stem cells, and their underlying molecular mechanisms have not yet been proven. Recently, studies have revealed that a reduced number of active stem/progenitor cells within endometrial tissue limits cyclic endometrial regeneration and subsequently decreases pregnancy success rates, suggesting that endometrial stem cells play a critical role in endometrial regeneration and subsequent endometrial receptivity. Methods We assessed whether aspirin treatment can inhibit various endometrial stem cell functions related to regenerative capacity, such as self-renewal, migration, pluripotency/stemness, and differentiation capacity, in vitro. Next, we evaluated whether SERPINB2 regulates the effects of aspirin on endometrial stem cell functions by depleting SERPINB2 expression with specific shRNA targeting SERPINB2. To further investigate whether aspirin also inhibits various endometrial stem cell functions in vivo, aspirin was administered daily to mice through intraperitoneal (i.p.) injection for 7 days. Results In addition to its previously identified roles, to the best of our knowledge, we found for the first time that acetylsalicylic acid directly inhibits various human endometrial stem cell functions related to regenerative capacity (i.e., self-renewal, migration, differentiation, and capacity) through its novel target gene SERPINB2 in vitro. Acetylsalicylic acid exerts its function by suppressing well-known prosurvival pathways, such as Akt and/or ERK1/2 signaling, through a SERPINB2 signaling cascade. Moreover, we also found that acetylsalicylic acid markedly inhibits regenerative capacity-related functions in endometrial stem cells within tissue. Conclusions We have found that acetylsalicylic acid has diverse effects on various endometrial stem cell functions related to regenerative capacity. Our findings are a critical step toward the development of more effective therapeutic strategies to increase the chances of successful pregnancy. Video Abstract Supplementary Information The online version contains supplementary material available at 10.1186/s12964-023-01339-2. Introduction Acetylsalicylic acid effectively inhibits platelet cyclooxygenases (COX-1 and 2) by irreversibly acetylating its active site (serine 530), thus blocking the formation of thromboxane A 2 [1], which in turn reduces platelet aggregation and subsequently enhances vasodilation.By these indirect mechanisms, acetylsalicylic acid may improve blood circulation in the uterine artery [2], and thus, some clinical researchers have recently tried to improve endometrial receptivity and subsequent implantation with consecutive acetylsalicylic acid treatment.However, there is ongoing controversy regarding its potential effects on endometrial receptivity and pregnancy success rates.For example, Zhang et al. found that the combined use of heparin and acetylsalicylic acid significantly reduces the miscarriage rate [3].In addition, lowdose acetylsalicylic acid therapy may delay or prevent the occurrence of preeclampsia during pregnancy [4].Chen et al. also showed that acetylsalicylic acid may increase endometrial thickness and regenerate endometrial damage after surgery for severe intrauterine adhesion [5].In contrast to these positive outcomes, meta-analyses, and randomized controlled trials (RCTs) regarding the use of low-dose acetylsalicylic acid in women undergoing in vitro fertilization (IVF) therapy to improve endometrial receptivity have shown controversial and ambiguous results [6,7].In addition, several studies focusing on the pregnancy outcomes of IVF patients have also obtained conflicting results [8].Similarly, the findings of Liping et al. suggest that there is no significant improvement in endometrial thickness or implantation rate with the use of acetylsalicylic acid treatment compared to control group [9].These conflicting results suggest that the impact of acetylsalicylic acid therapy on endometrial receptivity remains a subject of debate. The endometrium is a highly regenerative tissue that exhibits remarkable cyclic change from 7 to 14 mm during each cycle and is shed in the absence of implantation [10].Importantly, tissue-resident endometrial stem cells are responsible for the cyclic turnover of the endometrial layer in which the implantation takes place [11].Indeed, the continuous activation and recruitment of endometrial stem cells to sites of regeneration are necessary for achieving successful pregnancy [12].A recent study revealed that a decreased number of viable tissueresident stem cells significantly decreases endometrial receptivity and thus reduces pregnancy rates in patients with recurrent early pregnancy loss (RPL) [12].According to Lucas et al., a decreased population of clonogenic tissue-resident stem cells within the endometrium can lead to a chronic inflammatory response, potentially resulting in an increased incidence of RPL [13].Therefore, we have devoted our efforts to investigating whether acetylsalicylic acid has direct effects on various regenerative functions of endometrial stem cells, which play a critical role in endometrial growth and subsequent pregnancy outcomes, and thus affects endometrial receptivity. Importantly, we found for the first time that acetylsalicylic acid acts as a possible inhibitory factor of endometrial receptivity in that its application significantly inhibits various endometrial stem cell functions related to regenerative capacity, such as proliferative, migratory, and multilineage differentiation potential, in vitro and in vivo.In addition, we also found that acetylsalicylic acid, through its target gene SERPINB2 (also known as plasminogen activator inhibitor type 2), suppresses key pro-survival signaling, including ERK1/2 or Akt pathway, which are involved in essential stem cell functions, such as self-renewal capacity [14], recruitment [15], pluripotency [16], in many stem cell types.Indeed, SERPINB2 knockdown with specific shRNA markedly attenuated the acetylsalicylic acid-mediated inhibitory effects.Moreover, the activation of the PI3K/Akt or FAK/ERK1/2 pathways with selective activators significantly abolished the acetylsalicylic acid-mediated suppressive effects on endometrial stem cells.These results suggest that, in contrast to its previously reported beneficial effects on endometrial receptivity (e.g., enhancing uterine blood flow and tissue perfusion), acetylsalicylic acid suppresses tissue regeneration-associated functions of endometrial stem cells in vitro and in vivo as a possible inhibitory factor of endometrial receptivity. Isolation and culture of human endometrial stem cells Human endometrial stem cells were obtained with written informed consent from uterine fibroid patients and approved by the Gachon University Institutional Review Board (IRB No: GAIRB2018-134).The endometrial tissue was minced and digested in DMEM containing 10% FBS and 250 U/ml type I collagenase for 5 h at 37 °C in a rotating shaker, according to a previously established procedure [17].The resulting mixture was filtered through a 40 µm cell strainer to separate stromal-like stem cells from epithelial gland fragments and undigested tissue.Endometrial stem cells were isolated from other cell types using a single-density Percoll layer by centrifugation for 20 min at 1200 g.The isolated cells were washed twice in PBS and cultured in growth media consisting of various growth factors, including IGF, VEGF, EGF, basic FGF, hydrocortisone, ascorbic acid, heparin, and 10% FBS (Gibco BRL) at 37 °C in a humidified atmosphere of 5% CO 2 in air.After 3 days, colony-forming cells were isolated using cloning rings (Sigma-Aldrich). Isolation and culture of mouse uterine tissue-derived stem cells The isolation of stem cells derived from mouse uterine tissue was approved and conducted in accordance with the Institutional Animal Care and Use Committee (IACUC) (LCDI-2019-0169) of Gachon University.Uterine tissue was minced into small pieces and digested in DMEM containing 10% FBS and 250 U/ml type I collagenase for 5 h at 37 °C.The resulting mixture was filtered through a 40 µm cell strainer, and the isolated cells were cultured in EBM-2 medium (Lonza) with EGM-2 supplements at 37 °C and 5% CO 2 . Cell proliferation assay To determine the anti-proliferative capacity of acetylsalicylic acid, the MTT assay was performed following the manufacturer's protocol.Endometrial stem cells (2 × 10 4 cells/well) were seeded in 96-well plates and incubated for 24 h.The cells were then treated with increasing concentrations of acetylsalicylic acid for 48 h.The viable cells were measured at a wavelength of 570 nm using a Versa-Max microplate reader. In vitro cell migration assay We assessed the impact of acetylsalicylic acid on the migration ability of endometrial stem cells by calculating the ratio of cells that migrated in response to acetylsalicylic acid treatment versus the number of cells that migrated spontaneously.To track cell migration, we plated cells at a density of 1 × 10 5 cells/well in 200 μL of culture medium in the upper chambers of permeable Transwell supports (Corning Inc., Corning, NY, USA).These Transwell chambers had 8.0 μm pores in 6.5-mm diameter polycarbonate membranes and were arranged in a 24-well plate format.After incubation, non-migrated cells on the upper surface of each membrane were removed by scrubbing with laboratory paper.Migrated cells on the lower surface of each membrane were then fixed with 4% paraformaldehyde for 5 min and stained with hematoxylin for 15 min.Finally, we counted the number of migrated cells in three randomly selected fields of each well using a light microscope at 50X magnification.The difference between the groups was expressed as a fold change. Real-time PCR analysis Total RNA was extracted from human or mouse endometrial stem cells using commercial TRIzol ® reagent (Invitrogen Life Technologies, CA, USA) according to the manufacturer's recommended instructions.RNA purity was estimated by measuring the ratio of absorbance at 260 nm and 280 nm.The first-strand cDNA was synthesized using SuperScript II Reverse Transcriptase (Invitrogen Life Technologies, CA, USA) with 1 μg of total RNA.The first-strand cDNA was synthesized using Express SYBR-Green qPCR Supermix (BioPrince, Seoul, South Korea).qPCR was performed using a QIAGEN's realtime PCR cycler, the Rotor-Gene Q.The relative mRNA expression levels of the target genes were calculated as fold changes using the ΔΔCT method.The sequences of the PCR primers are listed in Table 1. Ingenuity pathway analysis The Ingenuity Pathway Analysis (IPA) version 2.0 software (Ingenuity Systems, Redwood City, CA) was used to analyze the genes related to SERPINB2.Genes that were differentially expressed between proliferative and non-proliferative cells (t-test, P < 0.005) were subjected to analysis of EGFR (GSE21618), IGFBP6 (GSE47856), M-CSF (GSE45630), PDGFRB (GSE116237), SCFR (GSE46045), or TGFß2 (GSE48990)-related genes.Fisher's exact test (P value) was used to measure the significance of each molecule, identifying differentially expressed genes from the microarray data that overlapped with known regulated genes.The activation score (z score) was used to indicate the predicted molecule status by comparing the observed differential regulation of genes ("up" or "down") in the microarray data relative to the literature-derived regulation direction, which can either activate or inhibit. Flow cytometry Flow cytometry analysis and cell sorting were conducted using FACS Calibur and FACS Aria machines (Becton Dickinson, Palo Alto, CA), respectively.Flow cytometry data were analyzed with FlowJo software (Tree Analysis of Gene Expression Omnibus (GEO) database Gene Expression Omnibus (GEO) is an open-access database repository that stores high-throughput gene expression data generated by genome hybridization arrays, chip sequencing, and DNA microarrays [18,19].Researchers can upload their experimental results in four categories: experimental designs, sample, platform, and raw data.Within each dataset, clinical or experimental samples are further classified into various experimental subgroups based on treatment, physiologic condition, and disease state.This categorized biological data is presented as a "GEO profile," which includes the dataset title, gene annotation, a chart showing the expression levels and rank for each gene across the samples [20].To analyze the expression profiles of EGFR, IGFBP6, M-CSF, PDGFRB, SCFR, or TGFß2 in response to various acetylsalicylic acid treatment conditions, we followed previously established procedures [20]. Adipocyte differentiation Human or mouse endometrial stem cells were cultured in low-glucose DMEM supplemented with 500 µM methylxanthine, 5 µg/mL insulin, and 10% FBS for 3 weeks with medium change twice per week.The formation of lipid droplets was confirmed by oil red O staining, and their abundance was measured by calculating the absorbance at 500 nm. Osteoblast differentiation Endometrial stem cells derived from either human or mouse were cultured in high-glucose DMEM supplemented with 0.1 µM dexamethasone, 10 mM β-glycerophosphate, 50 µM ascorbate, and 10% FBS for a period of three weeks with medium replacement twice per week.After differentiation, cells were stained with alizarin red S to detect the formation of new bone matrix. To quantify the presence of alizarin red S in the samples, the optical density (OD) of the solution was measured at 570 nm. Growth factor antibody array The manufacturer's protocol (Abnova AA0089) was followed to perform the assay.In brief, protein samples treated with acetylsalicylic acid or vehicle were incubated with antibody membranes overnight at 4 °C.After washing with wash buffer for 3 times, biotin-conjugated anticytokine antibodies were incubated with the membranes overnight at 4 °C.The membranes were then washed 3 times and incubated with HRP-conjugated streptavidin. Detection of signals of the growth factors that were spotted on the nitrocellulose membrane was performed using chemiluminescence. Evaluation of the effects of acetylsalicylic acid treatment in animal model Even though it does not entirely depict the physiological features of the human body, the outcomes achieved in vitro were validated through the use of mice, which are commonly utilized as in vivo models for evaluating efficacy.Animal experiments were conducted in accordance with the Institutional Animal Care and Use Committee (IACUC) (LCDI-2019-0169) of the Gachon University, and all protocols were approved.All experiments were designed and reported in accordance with the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.The animals were housed in standard cages with ad libitum access to food (standard chow diet) and water.The animal room was maintained on a 12-h light/ dark cycle at a constant temperature of 22 °C.The female mice (C57BL/6) were blindly and randomly divided into two groups (n = 5 per group sufficient to determine differences by treatment): control and acetylsalicylic acid treatment (50 mg/kg for 7 consecutive days intravenously).All animals were involved in data analyzing.After anesthesia and exsanguination by cardiac puncture, stem cells were isolated from uterine and adipose tissues.The isolated stem cells from endometrium, adipose tissue, or bone marrow were then cultured and expanded in vitro with continuous exposure to acetylsalicylic acid (2.5 mM) to simulate the physiological conditions of acetylsalicylic acid exposure in vivo.Then, the effects of aspirin on the self-renewal, migration, pluripotency, and differentiation capacity of stem cells in vivo were evaluated. Statistical analysis The experimental data were presented as mean ± standard deviation (SD) and were obtained from a minimum of three independent experiments.Statistical analysis among the experimental groups was performed using GraphPad Prism 5.0 (GraphPad Software, Inc., La Jolla, CA, US) with one-way ANOVA.Significance was set at p < 0.05. Acetylsalicylic acid significantly suppresses various endometrial stem cell functions related to regenerative capacity in vitro Human endometrial stem cells were isolated from endometrial biopsies according to the method described earlier by Cho et al. [21].These cells were characterized by analyzing various stem cell surface antigens using flow cytometry (Suppl.Figure 1A-B).Their pluripotency/ stemness was also assessed by inducing into adipogenic (Suppl.Figure 1C) and osteogenic (Suppl.Figure 1D) differentiation.A schematic representation of our results showing the novel functions of acetylsalicylic acid is presented in Fig. 1A.We assessed whether acetylsalicylic acid treatment can inhibit various endometrial stem cell functions related to regenerative capacity.The proliferative capacity of endometrial stem cells was markedly reduced by acetylsalicylic acid treatment (Fig. 1B).In addition, acetylsalicylic acid treatment also reduced endometrial stem cell migration (Fig. 1C).To further assess the acetylsalicylic acid-induced inhibitory effects on their migratory capacity, the protein levels of MMP-2 and -9, which induce the proteolytic degradation of the ECM and thus regulate cell migration, were analyzed (Fig. 1D).Interestingly, acetylsalicylic acid treatment significantly decreased their adipogenic (Fig. 1E) and osteogenic (Fig. 1F) differentiation capacities.The expression of major stemness-associated factors, including C-MYC, OCT4, NANOG, and SOX2, was also markedly reduced by acetylsalicylic acid (Fig. 1G).These results indicate that acetylsalicylic acid significantly suppresses various endometrial stem cell functions related to regenerative capacity, such as proliferative, migratory, stemness, and multilineage differentiation potential. The suppressive effects of acetylsalicylic acid on the regenerative capacity of endometrial stem cells are mediated through its target gene SERPINB2 Previous studies have found that toxic exposure may induce the premature senescence of resident stem cells in the endometrium [22], leading to adverse effects on endometrial functions [23,24].Recently, Part et al. revealed that SERPINB2 is involved in the regulation of various stem cell functions, such as proliferative, migratory capacity, and aging phenotypes [21].In this context, we therefore determined whether acetylsalicylic acid exerts its inhibitory effects on endometrial stem cells through SERPINB2 (Fig. 2A).Interestingly, our results revealed that acetylsalicylic acid treatment significantly increased SERPINB2 expressions in a dose-dependent manner (Fig. 2B).Consistently, daily intraperitoneal injections of acetylsalicylic acid (50 mg/kg) for 7 days suppressed SERPINB2 expressions in various tissue-derived stem cells, such as adipose tissue, endometrium, bone marrow (Fig. 2C).Thus, we evaluated whether SERPINB2 regulates the function of acetylsalicylic acid in endometrial stem cells by knocking down SERPINB2 expression using its specific shRNA (Suppl.Figure 2A-C).Indeed, the acetylsalicylic acid-mediated suppression of proliferative capacity was significantly attenuated by SERPINB2 depletion (Fig. 2D).The acetylsalicylic acid-induced suppression of migration potential (Fig. 2E) and the expressions of MMP-2 and -9 (Fig. 2F) were also clearly diminished by SERPINB2 knockdown.Importantly, we also observed that the acetylsalicylic acid-mediated suppression of adipogenic (Fig. 2G) and osteogenic (Fig. 2H) differentiation were significantly attenuated by SER-PINB2 depletion.In addition, the effects of acetylsalicylic acid on the expressions of several stemness-related transcription factors (OCT4, NANOG, and SOX2) were also abolished by its knockdown (Fig. 2I).These results indicate that SERPINB2 may act as a functional regulator to mediate the acetylsalicylic acid-induced inhibition of various endometrial stem cell functions related to regenerative capacity. The suppressive effects of acetylsalicylic acid on the regenerative capacity of endometrial stem cells of are mediated through the Akt and/or ERK1/2 pathways We then investigated possible molecular mechanisms underlying the acetylsalicylic acid-mediated inhibitory effects on the regenerative capacity of endometrial stem cells by assessing the effects of acetylsalicylic acid on the Akt and ERK1/2 pathways, which have previously been implicated in controlling the self-renewal [25], migratory capacities [26], and pluripotency/ stemness [27] of various stem cell types (Fig. 3A).We thus analyzed whether the Akt (Fig. 3B) or ERK1/2 (Fig. 3C) signaling were suppressed by acetylsalicylic acid using western blotting.We also analyzed signaling activities of PI3K/AKT and ERK1/2 pathways following treatment with acetylsalicylic acid at different concentrations (ranging from 0.1 mM to 5 mM).As a result, we observed a significant, dose-dependent decrease in the phosphorylation levels of the PI3K/ AKT (Suppl.Figure 3A and B) and ERK1/2 (Suppl.Figure 3C) pathways by acetylsalicylic acid treatment. We then assessed the effect of SERPINB2 knockdown on the acetylsalicylic acid-induced inhibition of the Akt or ERK1/2 signaling.Importantly, the inhibitory effects of acetylsalicylic acid on the Akt (Fig. 3D) or ERK1/2 (Fig. 3E) signaling were significantly abolished by SERPINB2 knockdown.These results indicate that SERPINB2 act as a potent regulator of the acetylsalicylic acid-induced Akt or ERK1/2 signaling cascades in endometrial stem cells.In addition, we evaluated the activation status of various Akt-or ERK1/2-related factors and their signaling networks to further analysis whether the activation of acetylsalicylic acid-associated signaling is negatively correlated with the Akt or ERK1/2 pathway using ingenuity pathway analysis (IPA).The expression of the negative regulators of several acetylsalicylic acid-related genes, including Akt1, CLCA2, and KIF24, were reduced in highly differentiated cells (Fig. 3F).In addition, the expression of the negative regulators of various acetylsalicylic acid-related genes, including MAPK1, MAPK2, and CDKN1A, were reduced in highly differentiated cells (Fig. 3G).Consistent with these results, the Gene Expression Omnibus (GEO) database indicated that the expressions of Akt, MAPK1, and MAPK3 are clearly reduced in various acetylsalicylic acid treatment conditions (Fig. 3H).In addition, we evaluated the activation of the Akt (Fig. 4A) or the ERK1/2 (Fig. 5A) signaling pathways on endometrial stem cell functions related to regenerative capacity, such as proliferative, migratory, stemness, and multilineage differentiation capacities, with or without acetylsalicylic acid treatment.Indeed, the acetylsalicylic acidinduced inhibitory effects on endometrial stem cell growth were clearly diminished by pretreatment with Akt (Fig. 4B) or ERK activations (Fig. 5B) with their specific activators.Akt activator SC79 (Fig. 4C-D) or ERK activator ceramide C6 (Fig. 5C-D) pretreatment significantly abolished the acetylsalicylic acid-meditated inhibitory effects on the migratory capacity and MMP-2/9 expression.The inhibitory effects on adipogenic and osteogenic differentiation were also clearly abolished by Akt (Fig. 4E-F) or ERK (Fig. 5E-F) activations.In addition, the effects of acetylsalicylic acid on the expression of stemness-associated genes, such as OCT4, NANOG, and SOX2, were also markedly attenuated by Akt (Fig. 4G) or ERK (Fig. 5G) activations.These findings highlight the potential significance of Akt and/or ERK1/2 signaling pathways in regulating various regenerative capacity associated endometrial stem cell functions through SERPINB2 signaling cascades.Endometrial stem cells were treated with or without acetylsalicylic acid (2.5 mM) for 72 h.The in vitro effects of acetylsalicylic acid treatment on the mRNA and protein expression levels of SERPINB2 in a dose-dependent manner were analyzed using real-time PCR and western blotting, respectively (B).In addition, mice were intraperitoneally treated with acetylsalicylic acid (50 mg/kg daily for 7 consecutive days) or vehicle (PBS).Various stem cell types were isolated from the mouse uterine endometrium, bone marrow, and adipose tissues, and the changes in SERPINB2 expression were subsequently analyzed using real-time PCR and western blotting (C).Endometrial stem cells were transfected with specific shRNA targeting SERPINB2 and then treated with or without acetylsalicylic acid (2.5 mM); the subsequent changes in self-renewal capacity were analyzed at 48 h using MTT assays (D).The effects of SERPINB2 knockdown on the acetylsalicylic acid-induced changes in the migration potential of endometrial stem cells were also assessed by in vitro Transwell assays (E) and western blotting with antibodies against MMP-2 and MMP-9 (F). The effects of SERPINB2 depletion on the acetylsalicylic acid-induced inhibition of adipogenic (G) and osteogenic (H) differentiation were evaluated using oil red O staining and alizarin red S staining, respectively.The attenuating effects of SERPINB2 depletion on the acetylsalicylic acid-induced changes in the levels of the pluripotency/stemness-associated transcription factors NANOG, OCT4, and SOX2 were also assessed by real-time PCR (I).β-actin was used as an internal control to normalize protein expression.PPIA was used as a reference gene to normalize gene expression.All experiments were performed in triplicates.Significant differences are presented.p < 0.05, p < 0.005, and *p < 0.001 (One-way ANOVA) Correlation analysis of acetylsalicylic acid-induced changes in the expression of multiple growth factors and various acetylsalicylic acid treatment conditions To investigate whether the suppressive effects of acetylsalicylic acid on various endometrial stem cell functions can be affected by the release of certain cytokines or growth factors, we performed growth factor antibody arrays using cells with or without acetylsalicylic acid treatment.Our analysis revealed significant alterations in the secretions of 40 proteins in acetylsalicylic acidtreated endometrial stem cells.Although other growth factors were shown to have only slight changes in their expression, the relative levels of nine prominent growth factors, namely, androgen receptor (AR); epidermal growth factor receptor (EGFR); glial cell-derived neurotrophic factor (GDNF); insulin-like growth factorbinding protein 6 (IGFBP-6); heparin-binding EGF-like growth factor (HB-EGF); macrophage colony-stimulating factor (M-CSF); platelet-derived growth factor receptor, beta polypeptide (PDGFRB); stem cell factor receptor (SCFR); and transforming growth factor 2 (TGFB2) were significantly reduced by acetylsalicylic acid treatment (Fig. 6A-B).These results indicate that these growth factors are probably responsible for the acetylsalicylic acidinduced suppression of the Akt or ERK1/2 signaling and its subsequent inhibition of regenerative capacities of endometrial stem cells.Furthermore, the GEO dataset also revealed that the expressions of the nine prominent growth factors detected are decreased in response to various acetylsalicylic acid treatment (Fig. 6C).We analyzed various signaling networks using the IPA software to further assess whether the nine growth factors detected are related to the signaling pathways controlling proliferative capacity.Positive regulators of EGFR, such as EGF, EGFR, and MAPK3, are activated in highly proliferating cells (Suppl.Figure 4A).Positive regulators of IGFBP6, such as FGF21 and PDGFBB, are activated in highly proliferating cells (Suppl.Figure 4B).Positive regulators of M-CSF2, such as FGFR2 and MAPK1, are activated in highly proliferating cells (Suppl.Figure 4C).Positive regulators of DGFRB, such as JUN and TGFB1, are activated in highly proliferating cells (Suppl.Figure 4D).Positive regulators of SCFR, such as HGF and VEGFA, are activated in highly proliferating cells (Suppl.Figure 4E).Positive regulators of TGFB2, such as AREG and FOS, are activated in highly proliferating cells (Suppl.Figure 4F).Taken together, these results indicate that these secreted proteins may be potent upstream regulators that affect the ERK1/2 or Akt signaling to mediate the effects of acetylsalicylic acid. Acetylsalicylic acid treatment inhibits various functions related to the regenerative capacities of tissue-resident endometrial stem cells in vivo Our in vitro data indicate that acetylsalicylic acid treatment may also suppress regenerative capacities associated functions of endometrial stem cells in vivo.Therefore, acetylsalicylic acid (50 mg/kg) was administered daily to mice through intraperitoneal (i.p.) injection for 7 days.Tissue-resident endometrial stem cells were then isolated from the mouse uterine tissue and cultured (Fig. 7A).Similar to our in vitro data, the results showed that acetylsalicylic acid treatment markedly reduced the self-renewal ability of endometrial stem cells (Fig. 7B).Transwell invasion assays also revealed the suppressive effects of acetylsalicylic acid treatment on the migration potential of endometrial stem cells in vivo (Fig. 7C).The acetylsalicylic acid-induced inhibition of endometrial stem cell migration was further assessed by western blot analysis using MMP-2/9 antibodies (Fig. 7D).In addition, acetylsalicylic acid treatment significantly decreased their adipogenic (Fig. 7E) and osteogenic (Fig. 7F) differentiation capacity in vivo.Real-time PCR results revealed that the levels of the stemness-associated transcription factors OCT4, NANOG, C-MYC, and SOX2 were markedly decreased by acetylsalicylic acid treatment (Fig. 7G).Importantly, we also investigated whether acetylsalicylic acid treatment could affect the regenerative capacity of the endometrium, which largely rely on the local endometrial stem cells in vivo. Our histological analysis showed that the thickness of the endometrial layer was clearly reduced by consecutive acetylsalicylic acid treatment (Fig. 7H).We further investigated whether acetylsalicylic acid treatment also suppresses regenerative capacities of other relevant types of stem cells, such as adipose tissue-derived stem cells (Suppl.Figure 5A) and bone marrow-derived stem cells (Suppl.Figure 6A) in vivo.Similar to endometrial stem cell results, acetylsalicylic acid treatment clearly reduced the self-renewal (Suppl.Figure 5B and 6B), migratory (Suppl.Figure 5C/D and 6C/D), and differentiation capacities (Suppl.Figure 5E/F and 6E/F) of these tissue-resident stem cells.Furthermore, the expressions of the stemness-associated genes NANOG, OCT4, C-MYC, and SOX2 were significantly decreased by acetylsalicylic acid treatment in both adipose tissue stem cells (Suppl.Figure 5G) and bone marrow stem cells (Suppl.Figure 6G) in vivo.Taken together, these findings suggest that consecutive acetylsalicylic acid treatment negatively affects the regenerative capacity of the endometrial stem cells by reducing their selfrenewal, migratory, and differentiation potentials. Discussion Intensive investigation of critical factors and cytokines that influence the quality of endometrial stem cells could lead to a better understanding of female infertility or repeated abortions that were previously unexplained.While there are various factors that can influence the functions of endometrial stem cells, the effects of acetylsalicylic acid have recently garnered attention due to its emerging regulatory functions in different disease conditions, such as cancers [28], gastric ulcers [29], hypertension [30], inflammation [31], and infertility [32].Acetylsalicylic acid (also known as Aspirin) is one of the most potent and widely used medicines ever discovered [33].Since its discovery more than 100 years ago [34], acetylsalicylic acid has been used clinically for its anti-inflammatory, anti-pyretic, and anti-nociceptive effects [35].The therapeutic mechanisms of acetylsalicylic acid have only been revealed in recent decades [36], beginning with its antiprostaglandin effects [37].Interestingly, the potential impact of acetylsalicylic acid on endometrial receptivity and subsequent pregnancy rates is still remains a topic of debate due to conflicting data: While some studies have shown positive effects of acetylsalicylic acid [38,39], others have found that it has no effect [40].Previous studies have examined the direct effects in endometrial cancer cell models in vitro [41].However, there is still much to investigate about the effects of acetylsalicylic acid on the various functions of normal endometrial cells and its underlying mechanisms involved.Recent research has raised new and challenging questions regarding the potential direct effect of acetylsalicylic acid on the regenerative capacities of endometrial stem cells, which play a critical role in endometrial regeneration and subsequent endometrial receptivity. In the present study, we obtained human endometrial stem cells from patients with uterine fibroids.However, we isolated these stem cells from unaffected normal tissue regions rather than uterine fibroid lesions.Therefore, we are confident that uterine fibroids do not affect the characteristics of these endometrial stem cells.Furthermore, as a result of analyzing the characteristics of isolated uterine stem cells using diverse cell surface markers (CD34, CD44, CD45, CD73, CD105, CD140b, CD146, and susD2), it becomes evident that these stem cells exhibit the typical characteristics of normal adult stem cells (Suppl.Figure 1B).Therefore, our findings suggest that uterine fibroids do not exert significant influence on the properties of isolated endometrial stem cells.In addition to expressing various surface markers, we induced the isolated endometrial stem cells to differentiate into adipocytes and osteoblasts to assess their normal differentiation abilities.As a result, we confirmed that the isolated endometrial stem cells indeed possessed the typical multilineage differentiation potential, demonstrated by their ability to differentiate into both adipocytes (Suppl.Figure 1C) and osteoblasts (Suppl.Figure 1D), with no significant impact from uterine fibroids.Previously, Cao et al. found that acetylsalicylic acid promotes the osteogenic differentiation of bone marrow mesenchymal stem cells (BM-MSCs) in vitro and in vivo, and BM-MSCs pretreated with acetylsalicylic acid have a better regenerative potential to repair calvarial bone defects in a swine model [42].Similarly, Tang showed that acetylsalicylic acid treatment significantly improves the immunomodulatory properties of BM-MSCs via the 15d-PGJ2/PPARγ/TGF-β1 pathway and that BM-MSCs pretreated with acetylsalicylic acid significantly ameliorate disease activity in vivo [43].In contrast to these positive outcomes, Hao et al. recently found that acetylsalicylic acid treatment significantly decreased the growth potential of BM-MSCs in a dose-dependent manner [44].Zhan et al. also observed that acetylsalicylic acid suppressed the adipogenic differentiation of BM-MSCs by disrupting epigenetic modifications in the cells [45].These results indicate that similar to its effects on endometrial receptivity, the effect of acetylsalicylic acid on various stem cell functions is still controversial, and the underlying mechanisms are not well understood.Furthermore, it is important to acknowledge that the dosage of acetylsalicylic acid utilized in this study exceeds clinical practice norms, with a dosage 2-3 times higher than typically administered.Moreover, the intravenous (IV) administration protocol eliminates the first-pass effect that occurs when the drug is taken orally, thereby maximizing absorption into local tissues [46].Consequently, it is reasonable to anticipate that a substantial quantity of doses can effectively reach the cells within the local tissue.Hence, it is plausible that the elevated concentration of acetylsalicylic acid within the local tissue may induce cell death, consequently diminishing various tissue regeneration-associated functions of endometrial stem cells, including self-renewal, migratory capacity, and pluripotency/stemness. In the present study, we found for the first time that acetylsalicylic acid treatment markedly suppressed multiple beneficial functions of endometrial stem cells, such as their proliferative, migration, and transdifferentiation capacities, both in vitro (Fig. 1A-G) and animal model (Fig. 7A-H), all of which are essential functions for endometrial receptivity [47].Importantly, the acetylsalicylic acid-induced suppression of endometrial stem cell growth (Fig. 2C), migration (Fig. 2D-E), multilineage differentiation (Fig. 2F-G) and pluripotency (Fig. 2H) were significantly attenuated by SERPINB2 depletion using specific shRNA.Previously, it has been reported that SERPINB2 levels are significantly increased in response to various differentiation-inducing agents in various cell types, such as leukemic cells [48], mononuclear cells [49], and keratinocytes [50], suggesting that SERPINB2 may be involved in the adverse effects of acetylsalicylic acid on stem cells as a potent downstream target of acetylsalicylic acid.Consistent with this model, our previous results also revealed that enhanced SERPINB2 levels markedly decreased the proliferation and pluripotency of stem cells [51].Cho et al. also found that stem cell senescence (aging) may enhance SERPINB2 expression, which in turn suppresses the anti-aging activity of Sonic hedgehog [21].In addition, other studies also found that increased SERPINB2 expression suppresses self-renewal ability and is associated with elevated levels of various differentiation-related markers [52].These findings suggest that SERPINB2 could play a crucial role in regulating the various effects of acetylsalicylic acid on endometrial stem cell functions.The diverse acetylsalicylic acid-induced effects were markedly attenuated by the activation of Akt (Fig. 4A-G) and ERK1/2 (Fig. 5A-G) signaling pathways.Moreover, we observed that the inhibitory effects of acetylsalicylic acid on the Akt and ERK1/2 signaling were markedly reduced upon SERPINB2 depletion.(Fig. 3D and E).These results indicate that the PI3K/Akt and FAK/ ERK1/2 signaling pathways may be involved in the acetylsalicylic acid-induced inhibition of various endometrial stem cell functions related to regenerative capacity as potent downstream modulators of SERPINB2 signaling cascades.While acetylsalicylic acid can directly influence various tissue regeneration-related functions, such as self-renewal, migratory capacity, differentiation potential, and pluripotency/stemness of endometrial stem cells, it is also plausible that its mechanism of action involves the regulation of several growth factors pivotal for stem cell function.Indeed, treatment with acetylsalicylic acid resulted in a significant reduction in the expression of growth factors such as AR, EGFR, GDNF, IGFBP-6, HB-EGF, M-CSF, PDGFRB, SCFR, and TGFB2, all of which are known to play crucial roles in governing stem cell function (Fig. 6A-B).It is postulated that these nine growth factors, whose expression was reduced by acetylsalicylic acid, may indirectly suppress the activity of the Akt and/or ERK1/2 signaling pathways.Therefore, to substantiate this hypothesis, further investigation is required, involving the overexpression of these nine growth factors either individually or in combination, followed by acetylsalicylic acid treatment, and subsequent assessment of Akt and/or ERK1/2 signaling pathway activities. Conclusion In conclusion, our study has revealed that acetylsalicylic acid exerts complex and varied effects on multiple functions of endometrial stem cells.However, the impact of these effects on endometrial receptivity and subsequent pregnancy outcomes remains unclear due to inconsistencies in previous studies.While some studies have reported beneficial effects of acetylsalicylic acid on endometrial receptivity and pregnancy outcomes, others have reported adverse effects.These inconsistencies may be explained by the differences in anatomy, genetics, and environmental factors.It is also widely recognized that animal models have limitations in accurately reflecting the physiological complexity of humans.Moreover, it remains unclear whether the effects of acetylsalicylic acid on endometrial stem cells extend to other types of endometrial constituting cells.Specifically, it is unknown whether acetylsalicylic acid exerts similar effects on endometrial stromal cells, glandular epithelial cells, and vascular endothelial cells.However, although there are still some unanswered questions, our findings may facilitate the development of more effective therapeutic strategies to increase pregnancy rates for patients with female infertility. Table 1 Primer sequences for quantitative RT-PCR Primer sequence TGC CAT CGC CAA GGA GTA G TGC ACA GAC GGT CAC TCA AA AAA GGC CCC CAA GGT AGT TA GCA CAA GAG TTC CGT AGC TG TGG GAT TTA CAG GCG TGA GC AAG CAA AGC CTC CCA ATC CC AGC CCT CAT TTC ACC AGG CC TGG GAC TCC TCC GGG TTT TG AAA TGG GAG GGG TGC AAA AGA GGA G CAG CTG TCA TTT GCT GTG GGT GAT G ACC CCC ATG ACT CCA GAG AACT GAG AGC GGA AGG ATG AAT GGAT NM_003106 F AAA TGG GAG GGG TGC AAA AGA GGA G GCC TAA GAT GAG CGC AAG TTG R CAG CTG TCA TTT GCT GTG GGT GAT G TAC TAG GCA GAT GGC CAC AGG 01143818 F ACC CCC ATG ACT CCA GAG AACT CGC ACA CAC AAC GTC TTG GA NM_003106 F AAA TGG GAG GGG TGC AAA AGA GGA G AGG ATG TAG GCG GTG GCT TT R CAG CTG TCA TTT GCT GTG GGT GAT G GCC TTA CGT ACA GTT GCA GC NM_001143818 F ACC CCC ATG ACT CCA GAG AACT GCC TTA CGT ACA GTT GCA GC GCA TTC AAA CTG AGG CAC CA AGC TTC TTT CCC CAT CCC A GAA GCG TGT ACT TAT CCT TCT TCA T GAG TGG AAA CTT TTG TCC GAGA NM_011443 F GAA GCG TGT ACT TAT CCT TCT TCA T ACT TAA TGG GCT TTA TCC TTTCC NM_011443 F GAA GCG TGT ACT TAT CCT TCT TCA T TGC GTC CTC AAT CTC ATC G R GAG TGG AAA CTT TTG TCC GAGA Fig. 1 Fig.1Acetylsalicylic acid treatment markedly suppresses multiple beneficial functions of endometrial stem cells in vitro.We evaluated whether acetylsalicylic acid treatment inhibits various regenerative capacity-associated functions of endometrial stem cells, such as self-renewal, migration, pluripotency and multilineage differentiation, in vitro (A).The inhibition of self-renewal capacity by treatment with multiple concentrations of acetylsalicylic acid (100 nM, 500 nM, 1 mM, 2.5 mM, and 5 mM) was analyzed at 48 h using MTT assays.The stem cell proliferation rates (%) were assessed by representing the viability of the acetylsalicylic acid-treated cells as a percentage of the viability of the vehicle-treated cells (B).Endometrial stem cells were treated with acetylsalicylic acid (2.5 mM) for 72 h, and the acetylsalicylic acid-induced inhibition of migratory potential was then analyzed by Transwell migration/invasion assays (C).The levels of the well-known migration regulatory proteins MMP-2 (72 kDa) and MMP-9 (92 kDa) in cells with or without acetylsalicylic acid treatment were analyzed using western blotting (D).Endometrial stem cells were incubated in adipogenic or osteogenic differentiation medium for 2 weeks with or without acetylsalicylic acid (2.5 mM) treatment.The acetylsalicylic acid-induced inhibition of the adipogenesis (E) and osteogenesis (F) of endometrial stem cells was analyzed using oil red O and alizarin red S staining, respectively.The cytoplasmic calcium concentration and lipid droplet (LD) formation within differentiated cells were assessed by measuring the absorbance values of the solubilized cells at wavelengths of 500 nm and 570 nm, respectively (E).The acetylsalicylic acid-induced inhibition of the expression of various pluripotency/stemness markers (C-MYC, NANOG, OCT4, and SOX2) was evaluated using real-time PCR (G).β-actin was used as an internal control to normalize protein expression.PPIA was used as a reference gene to normalize gene expression.All experiments were performed in triplicates.Significant differences are presented.p < 0.05, p < 0.005, and p < 0.001 (One-way ANOVA) Fig. 2 Fig.2Acetylsalicylic acid exerts diverse effects on the regenerative capacity of endometrial stem cells through its target gene SERPINB2.A schematic representation of the regulatory role of SERPINB2 in mediating the diverse effects induced by acetylsalicylic acid is shown (A).Endometrial stem cells were treated with or without acetylsalicylic acid (2.5 mM) for 72 h.The in vitro effects of acetylsalicylic acid treatment on the mRNA and protein expression levels of SERPINB2 in a dose-dependent manner were analyzed using real-time PCR and western blotting, respectively (B).In addition, mice were intraperitoneally treated with acetylsalicylic acid (50 mg/kg daily for 7 consecutive days) or vehicle (PBS).Various stem cell types were isolated from the mouse uterine endometrium, bone marrow, and adipose tissues, and the changes in SERPINB2 expression were subsequently analyzed using real-time PCR and western blotting (C).Endometrial stem cells were transfected with specific shRNA targeting SERPINB2 and then treated with or without acetylsalicylic acid (2.5 mM); the subsequent changes in self-renewal capacity were analyzed at 48 h using MTT assays (D).The effects of SERPINB2 knockdown on the acetylsalicylic acid-induced changes in the migration potential of endometrial stem cells were also assessed by in vitro Transwell assays (E) and western blotting with antibodies against MMP-2 and MMP-9 (F).The effects of SERPINB2 depletion on the acetylsalicylic acid-induced inhibition of adipogenic (G) and osteogenic (H) differentiation were evaluated using oil red O staining and alizarin red S staining, respectively.The attenuating effects of SERPINB2 depletion on the acetylsalicylic acid-induced changes in the levels of the pluripotency/stemness-associated transcription factors NANOG, OCT4, and SOX2 were also assessed by real-time PCR (I).β-actin was used as an internal control to normalize protein expression.PPIA was used as a reference gene to normalize gene expression.All experiments were performed in triplicates.Significant differences are presented.p < 0.05, p < 0.005, and p < 0.001 (One-way ANOVA) Fig. 3 Fig.3The attenuating effect of SERPINB2 knockdown on the acetylsalicylic acid-induced suppression of the Akt or ERK1/2 signaling cascade.A schematic representation of the role of SERPINB2 as an upstream regulator of the acetylsalicylic acid-induced PI3K/Akt and FAK/ERK1/2 signaling cascades is shown (A).Endometrial stem cells were treated with or without acetylsalicylic acid (2.5 mM) for 15 min, and the subsequent changes in the phosphorylation (activation) levels of signaling molecules (i.e., Akt, PI3K, FAK, and ERK1/2) were assessed by western blotting (B-C).Endometrial stem cells were treated with acetylsalicylic acid (2.5 mM) alone or concomitantly transfected with a specific shRNA targeting SERPINB2; the subsequent changes in the phosphorylation states of these signaling molecules were evaluated by western blotting (D-E).The activation states (either activated or inactivated) of various Akt1 (GSE100752) (F) or MAPK1/3 (ERK1/3) (GSE129144/GSE76381) (G)-associated genes/ transcription factors in proliferating and nonproliferating cells were analyzed using the ingenuity pathway analysis (IPA) software.The GEO metadata were also analyzed to further investigate the contributions of the Akt and MAPK1/3 (ERK1/3) signaling pathways to various acetylsalicylic acid treatment conditions (H).β-actin was used as an internal control to normalize protein expression.All experiments were performed in triplicates.Significant differences are presented.p < 0.05, p < 0.005, and p < 0.001 (One-way ANOVA) Fig. 4 Fig.4 Activation of the Akt signaling cascade alleviates the acetylsalicylic acid-induced suppression of various endometrial stem cell functions related to regenerative capacity.A schematic representation of the regulatory role of the PI3K/Akt signaling cascade in mediating the effects of acetylsalicylic acid is shown (A).Endometrial stem cells were pretreated with the specific Akt activator SC79 (10 µM) for 1 h and then treated with acetylsalicylic acid (2.5 mM) for 48 h, and the subsequent effects of acetylsalicylic acid on self-renewal capacity were assessed using MTT assays.The stem cell proliferation rates (%) were assessed by representing the viability of the acetylsalicylic acid-treated cells as a percentage of the viability of the vehicle-treated cells (B).The effect of the Akt activator on the acetylsalicylic acid-induced change in migration potential was evaluated by in vitro Transwell assays (C) and western blotting with antibodies against MMP-2 and MMP-9 (D).Endometrial stem cells were pretreated with 10 µM SC79 for 1 h and then treated with 2.5 mM acetylsalicylic acid for 48, and the subsequent changes in adipogenic (E) and osteogenic (F) differentiation were analyzed by oil red O and alizarin red staining, respectively.The cytoplasmic calcium concentration and lipid droplet (LD) formation within differentiated cells were assessed by measuring the absorbance values of the solubilized cells at wavelengths of 500 nm and 570 nm, respectively.The effect of 10 µM SC79 on the acetylsalicylic acid-induced inhibition of the pluripotency-associated transcription factors NANOG, OCT4, and SOX2 was measured by real-time PCR (G).β-actin was used as an internal control to normalize protein expression.PPIA was used as a reference gene to normalize gene expression.All experiments were performed in triplicates.Significant differences are presented.p < 0.05, p < 0.005, and p < 0.001 (One-way ANOVA) Fig. 5 Fig.5 Activation of the ERK1/2 signaling cascade alleviates the acetylsalicylic acid-induced suppression of various endometrial stem cell functions related to regenerative capacity.A schematic representation of the regulatory role of the FAK/ERK1/2 signaling cascade in mediating the effects of acetylsalicylic acid is shown (A).Endometrial stem cells were pretreated with the specific ERK1/2 activator ceramide C6 (10 µM) for 1 h and then treated with acetylsalicylic acid (2.5 mM) for 48 h, and the subsequent effects of acetylsalicylic acid on self-renewal capacity were assessed by MTT assays.The stem cell proliferation rates (%) were assessed by representing the viability of the acetylsalicylic acid-treated cells as a percentage of the viability of the vehicle-treated cells (B).The effect of the ERK1/2 activator on the acetylsalicylic acid-induced changes in migratory capacity was evaluated by in vitro Transwell assays (C) and western blotting with antibodies against MMP-2 and MMP-9 (D).Endometrial stem cells were pretreated with 10 µM ERK1/2 activator ceramide C6 for 1 h and then treated with 2.5 mM treatment for 48, and the subsequent effects of acetylsalicylic acid on adipogenic (E) and osteogenic (F) differentiation were evaluated by oil red O and alizarin red staining.The cytoplasmic calcium concentration and lipid droplet (LD) formation within differentiated cells were assessed by measuring the absorbance values of the solubilized cells at wavelengths of 500 nm and 570 nm, respectively.The effect of 10 µM ceramide C6 on the acetylsalicylic acid -induced inhibition of the pluripotency-associated transcription factors NANOG, OCT4, and SOX2 was analyzed by real-time PCR (G).β-actin was used as an internal control to normalize protein expression.PPIA was used as a reference gene to normalize gene expression.All experiments were performed in triplicates.Significant differences are presented.p < 0.05, p < 0.005, and p < 0.001 (One-way ANOVA) Fig. 6 Fig. 7 Fig.6 Acetylsalicylic acid treatment reduces the secretion of multiple cytokines or growth factors related to the regeneration-associated gene network in vitro.The human growth factor/cytokine antibody array was analyzed using acetylsalicylic acid-treated or nontreated protein samples.The nitrocellulose membrane was spotted with specific antibodies for 40 different cytokines, growth factors, and receptors.Nine growth factors (AR, EGFR, GDNF, HB-EGF, IGFBP-6, M-CSF, PDGFRB, SCFR, and TGFß2) were clearly reduced in the protein samples from the acetylsalicylic acid-treated cells (A-B).The GEO metadata were also analyzed to further investigate the correlations between acetylsalicylic acid treatment and the nine downregulated proteins (C).All experiments were performed in triplicates.Significant differences are presented.p < 0.05, p < 0.005, and *p < 0.001 (One-way ANOVA)	2023-11-10T14:54:48.177Z	2023-11-10T00:00:00.000	{ "year": 2023, "sha1": "5f4fb0b168b8864faabec3035df1c3178f9a8373", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Springer", "pdf_hash": "5f4fb0b168b8864faabec3035df1c3178f9a8373", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
97188234	pes2o/s2orc	v3-fos-license	Refractrometic Studies of Binary Mixtures of Flavoured Compounds and Some Hydrocarbons Abstract: Refractive indices have been measured for the binary liquid mixtures of n-butylethanoate and 3methylethanoate with cyclohexane, benzene, 1,4-dimethylbenzene and 1,3,5-trimethylbenzene over the entire composition range at 308.15K. From the experimental data, the values of deviation in refractive index ( n), molar refraction (R) and deviation in molar refraction ( R) have been determined. The values of n and R have been fitted to the Redlich Kister polynomial equation to determine the standard deviation. INTRODUCTION Physical properties of several esters in their binary mixtures have been already reported [1][2][3][4][5][6][7][8].However, studies in this regard for the determination of refractive indices of the binary mixtures of n-butylethanoate and 3-methylbutylethanoate at 308.15K are still lacking.Esters are an important class of organic compounds having characteristic fruity odour.They are used as component of perfumes and flavourings.Higher esters have low solubility in water and used as extraction solvents for fine chemicals particularly for certain antibiotics.They are also useful in cosmetics and materials for personal care. 3-methylbutylethanoate is used as solvent for old oil colors, perfuming, shoe polish, metallic paints and in photographic films. In our laboratory refractive index studies have been made on binary systems of bromoalkanes and cyclohexane and aromatic hydrocarbons.Literature reveals that binary systems of esters and aromatic hydrocarbons have rarely been studied employing refractive index measurements.In view of this, binary systems of flavoured compounds viz.1-butylethanoate and 3-methylbutylethanoate and aromatic hydrocarbons viz.benzene, 1,4-dimethylbenzene and 1,3,5-trimethylbenzene having successive methylation and differing in the polarisablity have been selected in the present investigation.Since cyclohexane has a six membered ring without any -electrons, binary mixtures of both the esters with it are also studied at experimental temperature in order to have a reference point for composition of unlike molecular interactions. Binary mixtures were prepared by weight covering whole mole fractions range in a sealed glass vials and component liquids were injected by means of syringe to minimize evaporation losses during sample preparation.Weights were measured employing single pan analytical balance (Model K-15 Deluxe, K. Roy Instruments Pvt. Ltd.Varanasi) with a precision of ± 0.00001x10 -3 kg. Densities of the liquids and their binary mixtures were measured employing stem pyknometer having bulb capacity about 12.0 x10 1 and 2 respectively.Measurements of refractive indices were carried out after attainment of constant temperature by the sample.An average of three to four measurements for each sample was taken. RESULTS AND DISCUSSION The experimental values of densities and refractive indices at 308.15 K for all the four binary mixtures and Where n and n 1 , n 2 are refractive indices of the mixtures and the components 1 and 2 are refractive indices of the mixtures and the components 1 and 2 respectively are also recorded in Tables 1 and 2. Redlich-Kister equation [11] given below Where A, B and C are the coefficients of the equation is generally used by the chemists to analyse thermodynamic data on liquid mixtures.However, Desnoyers and Perron [12] has suggested in agreement with the original statements of Redlich and Kister, that plot of Y E /x 1 (1-x 1 ) is better for this purpose. In the present investigation Redlich -Kister equation was fitted to the evaluated n values and the coefficients of equation and standard deviation ( n) for all the systems are recorded in Tables (3)(4). Perusal of Tables 1 and 2 reveals that refractive indices for all the mixtures increases with mole fractions of hydrocarbons.It shows that the order of refractive index for binary mixtures of n-butylethanoate and 3-methylbutylethanoate with hydrocarbons is.Cyclohexane< benzene < 1,4-dimethylbenzene< 1,3,5-trimethylbenzene And this order is shown in Figures (3)(4).Experimental n values are used to evaluate molar refraction, R for all the binary mixtures studied by equation: Variations of deviation in refractive indices, n for the mixtures studied with concentrations of Where R is the molar refraction, x 1 and x 2 are mole fractions of components 1 and 2 M 1 and M 2 are the molecular weight of component 1 and 2, is the density, n is the refractive index. R values for all the mixtures also given in Tables 1 and 2 for both the systems with hydrocarbons variations of R with composition of mixtures are linear for all the binary mixtures.Values of R at equimolar compositions follow the trend similar that of n values as given above and shown in Figures 7 and 8. Deviation in molar refraction from the linear blending rule ( R) have been defined on mole fraction basis [13,14].Considering molar refraction as isomorphic to the molar volume.However, others [15,16] have defined it on volume fraction basis.In the present investigation R has been evaluated on volume fraction basis as equation 5: Where R 1 , R 2 are the molar refractions of pure components and 1 and 2 are their volume fractions in binary mixtures.The R values with composition for all the systems at experimental temperature evaluated by equation ( 4) are recorded in Tables 1 and 2 It is evident from Figures 5 and 6 that R values are negative for binary systems of n-butylethanoate with all the hydrocarbons.But for the binary systems of 3methylbutylethanoate R values are negative for cyclohexane and benzene positive for 1,4dimethylbenzene and 1,3,5-trimethylbenzene. CONCLUSION Refractive index and density of binary mixtures are useful for evaluation of molar refraction.These evaluations give information about molecular interactions between the components of the mixture. Figure 1 : Figure 1: Operating Principle of the Abbe Refractometer. Temperature of the sample was kept constant by circulating water from thermostat at constant experimental temperature with the help of tullu pump through the upper and lower prism boxes of the Abbe Refractometer.The diagram and photograph of the refractometer is shown in Figures	2019-04-06T13:07:01.922Z	2013-03-01T00:00:00.000	{ "year": 2013, "sha1": "444a6c1332e29c244bd7f02146c0c4d1b53c2cb0", "oa_license": "CCBY", "oa_url": "https://lifescienceglobal.com/pms/index.php/JASCM/article/download/560/pdf", "oa_status": "GREEN", "pdf_src": "Anansi", "pdf_hash": "444a6c1332e29c244bd7f02146c0c4d1b53c2cb0", "s2fieldsofstudy": [ "Chemistry" ], "extfieldsofstudy": [ "Chemistry" ] }
196812556	pes2o/s2orc	v3-fos-license	Acute and chronic hypoxia differentially predispose lungs for metastases Oscillations in oxygen levels affect malignant cell growth, survival, and metastasis, but also somatic cell behaviour. In this work, we studied the effect of the differential expression of the two primary hypoxia inducible transcription factor isoforms, HIF-1α and HIF-2α, and pulmonary hypoxia to investigate how the hypoxia response of the vascular endothelium remodels the lung pre-metastatic niche. Molecular responses to acute versus chronic tissue hypoxia have been proposed to involve dynamic HIF stabilization, but the downstream consequences and the extent to which differential lengths of exposure to hypoxia can affect HIF-isoform activation and secondary organ pre-disposition for metastasis is unknown. We used primary pulmonary endothelial cells and mouse models with pulmonary endothelium-specific deletion of HIF-1α or HIF-2α, to characterise their roles in vascular integrity, inflammation and metastatic take after acute and chronic hypoxia. We found that acute hypoxic response results in increased lung metastatic tumours, caused by HIF-1α-dependent endothelial cell death and increased microvascular permeability, in turn facilitating extravasation. This is potentiated by the recruitment and retention of specific myeloid cells that further support a pro-metastatic environment. We also found that chronic hypoxia delays tumour growth to levels similar to those seen in normoxia, and in a HIF-2α-specific fashion, correlating with increased endothelial cell viability and vascular integrity. Deletion of endothelial HIF-2α rendered the lung environment more vulnerable to tumour cell seeding and growth. These results demonstrate that the nature of the hypoxic challenge strongly influences the nature of the endothelial cell response, and affects critical parameters of the pulmonary microenvironment, significantly impacting metastatic burden. Additionally, this work establishes endothelial cells as important players in lung remodelling and metastatic progression. (A) eletion of HIF-1 in primary C carrying double-floxed HIF-1 allele and exposed to Cre-expressing Adenovirus is consistently above 0% (WT control cells were treated with adenovirus expressing β-galactosidase); (B) epresentative IF images of HIF-isoform signal in all treatments and all genotypes; Animals expressing lung endothelium specific L1 Cre show visibly less HIF signal for the respective isoform. Original whole membrane images of WB using nuclear extracts of normoxic and hypoxic primary cells probed for HIF-1 , cropped and shown in main text. PV F membranes were cut at 5 a (using Bio ad Precision Plus dual collor ladder). Bottom halft was used to probe for -actin or TATA-binding protein, and upper half for HIF-1 . Multiple membranes were occasionally scanned at once when target signal intensity was comparable. ections framed in blue represent for HIF-1 and green -actin control. Whole scans of western blot of nuclear extracts of primary cells probed for HIF-2 , cropped and shown in main text (biological replicate 2). PV F membranes were cut at 5 a (using Bio ad Precision Plus dual colour ladder). Bottom half was used to probe for -actin and upper half for HIF-2 . Multiple membranes were occasionally scanned at once when target signal intensity was comparable. ections framed in orange represent for HIF-2 and green -actin control, in the whole scan images. α α β− cans of western blot membranes of whole lung protein probed for HIF-1 , HIF-2 and -actin control, cropped and shown in the main text. PV F membranes were cut at 5 a (using Bio ad Precision Plus dual colour ladder). amples were prepared in one large mix and same volume (corresponding to 15ug of protein) were loaded in multiple wells, to probe for multiple targets. The upper half of the membranes was used for targets running above 5 a, and the bottom half was used to probe for -actin and other targets smaller than 5 a. Cropped sections used in main gure are framed in blue for HIF-1 , orange for HIF-2 and green for -actin control. Images of the membranes are provided. These were superimposed to the luminescence images to locate the targets in relation to the standards. The membrane used for HIF-2 probe was also used for i O (shown in upplementary g. 2 ). Extended methods -Reiterer et al., Acute and chronic hypoxia differentially predispose lungs for metastases Animal models: Deletion of HIF-1α or HIF-2α in lung EC was obtained by crossing female animals homozygous for the floxed alleles of HIF-1α 1 or HIF-2α 2 (double-floxed, DF) to HIF-1α df /L1Cre + or HIF-2α df /L1Cre + males 3 . Experimental cohorts including L1Cre + were perfomed using Cre-littermate controls (DF animals with no deletion, physiologically wildtype, as previously shown 1,2 ). We have previously shown that L1-driven Cre expression is confined to the pulmonary endothelium 4 . C57Bl/6 WT mice were purchased from the Charles River laboratories (UK). In vivo Hypoxia treatments: Eight-week old male animals were exposed to 10% O2 for either 24 h (acute exposure, optimised for preferential activation of HIF-1 in lung tissue) or 10 d Metastasis Assay: A total of 5 × 10 5 LLC cells in 200 μl sterile PBS were injected into tail veins of animals pre-conditioned in hypoxia or normal air controls. Injections were performed inside the hypoxia chamber for the hypoxic pre-conditioned animals, and animals were that all animals were exactly the same age at the time of injection (hypoxia treatments started at day -10 for chronic treatments, day -1 for acute treatments, and at day 0 all animals received tumour cells from the same culture batch). Animals were transferred to normal air after tumour cell injection. Lung tissue was collected either 24h (to monitor immediate responses) or 14 days post-injection (to quantify metastatic take). Lung tissue was collected directly into OCT, immediately processed for flow cytometry, snap frozen in liquid N2 (for protein and RNA extractions), or perfused with 10 mg/mL heparin in PBS, and fixed in 4% PFA for paraffin embedding and H&E staining. Lung tumors were counted in H&E-stained 10 μm evenly spaced sections across whole lungs (≥ 20 sections per animal). Tumor area was quantified using Image J software 5 . Click-iT reaction buffer and additive, according to manufacturer's protocol. Secondary immunostaining, when performed, was done after this step. All slides were mounted in Vectashield anti-fade mounting medium with DAPI (Vector) and imaged the following day. Microscopy and image analysis: Images were taken with Leica fluorescence microscope using a x20, x40 or x63 (oil immersion) objective. Gain and offset were set to negative controls, and used to standardize image acquisition for each experiment (tissue: three images per section, a ≥ 8 randomly chosen sections from each animal; cells: ≥ 4 images per slide chamber, three slides per treatment). Quantification of fluorescence was done using ImageJ. After setting the threshold to a duplicate image, stained areas are identified and selected. "Dark background" is selected for fluorescence. "Analyze-Analyze particles" was used for measurement of various parameters; background values were acquired by selecting three random areas in gray-scale image using the circle tool and "Analyze-Measure" function. Quantification in tissue samples was performed as follows: The mean fluorescence for each image was calculated: Extended methods -Reiterer et al., Acute and chronic hypoxia differentially predispose lungs for metastases To correct for uneven fluorescence the following formula was used: The number of TUNEL + cells was identified by DAPI co-staining. Confocal Imaging and co-localization: Images were acquired in a Leica SP5 confocal microscope using x20, x40 or x63 (Oil Immersion) objective. Images were acquired through LAS (Leica) and analysed using ImageJ 5 . Pearson's correlation was calculated to assess co-localization through the Coloc 2 plugin. Cell culture and hypoxia treatments: Primary EC were isolated and cultured from lungs of HIF-1α df , HIF-2α df , iNOS df male animals between 6 and 8 weeks of age, as previously described 12 . Cre recombinase-mediated gene deletion was performed ex-vivo by overnight Western blotting: Equal amounts of protein (15 μg) were obtained from either cells or whole lung tissue lysed in RIPA buffer, resolved in 3-8% acrylamide Tris-Acetate gels (Life Sciences, EA0375BOX) and transferred to PVDF membrane using BioRad Transblot Turbo Transfer System. Primary antibodies were used at 1:1,000 (iNOS, sc-651; HIF-1α, NB-100-049) or 1:500 dilutions (HIF-2α, R&D AF2997). Membranes were cut at 75KDa band and targets were probed on the upper half of the membrane, whereas -actin (Sigma, A1978, running at ~50KDa), used as normalization control, was probed on the bottom half of the membrane. Membranes were not re-probed for multiple antibodies. Protein signals were detected following secondary incubation with HRP-conjugated antibodies for 1h at room temperature, and ECL Plus chemiluminescence detection kit (Amersham, Cat. # RPN2232), according to manufacturer's protocol. Image capture and quantification were performed	2019-07-17T13:03:22.074Z	2019-07-15T00:00:00.000	{ "year": 2019, "sha1": "414067aa084054b611b2040e4134e8038e29372b", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41598-019-46763-y.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "cc0dca197c2de69fb3e2ae43fa81d131b0247160", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
38338405	pes2o/s2orc	v3-fos-license	Reorientation mechanisms of block copolymer / CdSe quantum dot composites under application of an electric field † Timeand temperature-resolved in situ birefringence measurements were applied to analyze the effect of nanoparticles on the electric field-induced alignment of a microphase separated solution of poly(styrene)-block-poly(isoprene) in toluene. Through the incorporation of isoprene-confined CdSe quantum dots the reorientation behavior is altered. Particle loading lowers the order–disorder transition temperature, and increases the defect density, favoring nucleation and growth as an alignment mechanism over rotation of grains. The temperature dependent alteration in the reorientation mechanism is analyzed via a combination of birefringence and synchrotron SAXS. The detailed understanding of the effect of nanoparticles on the reorientation mechanism is an important prerequisite for optimization of electricfield-induced alignment of block copolymer/nanoparticle composites where the block copolymer guides the nanoparticle self-assembly into anisotropic structures. Introduction Approaching the nanometer length scale, classical photo lithographic methods reach their limits, and self-assembly processes are becoming a promising and cost-efficient alternative.In particular, block copolymer/nanoparticle composite materials have gained considerable interest in the past decade due to their wide variety of potential applications.While the nanoparticles exhibit interesting catalytic, 1 optical, 2-4 mechanical, 3 or magnetic properties, 5 block copolymers, can effectively control particle location and patterns. 6,7Utilizations include photonic band gap materials, 7,8 solar cells, 9 and catalytic 1 and biomedical devices. 10he properties of the composite materials are not only determined by particle size and shape, but also by the spatial distribution within the polymer matrix.Therefore, precise control over the particle assembly and orientation is required.Block copolymers, which are capable of forming a rich variety of structures in a size range of 10-100 nm, are the ideal scaffold for the assembly.Through selective insertion into one block copolymer domain, highly ordered arrays of nanoparticles are producible.The crucial parameter determining the affinity to the individual block copolymer constituents is the chemical surface modification of the particle. 11The localization within the domain is mainly influenced by its core diameter and the grafting densities and molecular weights of the ligands. 12f no directing patterns or orienting stimuli are applied, the block copolymers form an isotropic multidomain structure.However, for most applications precise control over orientation with a minimum amount of defects is favorable.Different techniques have been established to obtain highly aligned samples from block copolymers, for example the application of external stimuli such as magnetic 13,14 or electric fields, 15 shear force, 16,17 temperature gradients, 18 and patterned substrates [19][20][21] have been thoroughly investigated. Foundation of the realignment process under application of an electric field is the difference in dielectric permittivity De = e A À e B between the copolymer blocks.Dielectric interfaces perpendicular to the electric field vector are electrostatically unfavorable compared to those parallel to the external field.The energy difference between the two orientations is proportional to the second power of the dielectric contrast (De) 2 / e between the two blocks devided by the mean dielectric permittivity of the sample. 22To obtain the energetically favorable orientation with interfaces parallel to the electric field vector, block copolymers can undergo three different microscopic mechanisms of alignment: nucleation and growth (NG), rotation of grains (RG), 23,24 and selective disordering (SD).Which mechanism is exploited strongly depends on the temperature and preorientation of the sample.Selective disordering is merely found in close vicinity to the order-disorder transition temperature T ODT and therefore will not be discussed in this publication.For a detailed explanation the authors refer to an article by Ruppel et al. 25 A precise description of the two other reorientation mechanisms will be given in the course of this publication. Although much effort has been given to understand the effects of electric fields on the block copolymer microstructure, [26][27][28][29] only few studies exist on how nanoparticles influence the electric fieldinduced alignment.Cell dynamic system simulations by Yan and coworkers revealed that the nanoparticles alter the morphology and reorientation dynamics of block copolymers under the influence of electric fields. 30Furthermore, Yan et al. anticipated that the inclusion of nanoparticles leads to an alteration in alignment mechanism with a preference of NG over RG. 31 Liedel et al. showed that selectively-confined gold nanoparticles lower the critical field strength necessary to align poly(styrene)-block-poly(2-vinylpyridine) thin films. 32n this publication the electric field-driven alignment of microphase separated polystyrene-block-polyisoprene solutions with polyisoprene-confined oleylamine-capped CdSe nanoparticles (CdSe-Np) is investigated by means of a combination of birefringence and in situ synchrotron small-angle X-ray scattering.][35] Chen et al. proposed a combination of in situ rheo-optical measurements, ex situ electron microscopy and in situ SAXS to probe alterations in microstructure and orientation upon flow-induced alignment. 36Birefringence measurements offer excellent time resolution while providing information on the degree of alignment of optically anisotropic block copolymer microstructures and are therefore well suitable for the real-time analysis of the alignment kinetics. We demonstrate that nanoparticles are capable of switching the reorientation mechanism of electric field-induced alignment as anticipated by Yan and coworkers.Poly(styrene)-block-poly-(isoprene) (SI) (33.2 wt%) in toluene was chosen as a model system due to the fact that its alignment kinetics and reorientation mechanisms under exposure to an electric field have been intensively analyzed via synchrotron small angle X-ray scattering (SAXS). 23,37Therefore, the comparison to previously performed experiments is straightforward.Oleylamine functionalized CdSe quantum dots were incorporated into the lamella-forming block copolymer.Previous work with polystyrene-block-polyisoprene nanofibers revealed that the oleylligands of the particles are selectively confined within the polyisoprene phase. 38Materials and methods Sample preparation Lamella forming poly(styrene)-block-poly(isoprene) of a molecular weight of 108 kg mol À1 and a polydispersity index of 1.05 was synthesized by sequential living anionic polymerization.The volume fraction of the poly(styrene) block amounts 0.42 (S 42 I 58 108 ).33.2 wt% solutions of the block copolymer in the non-selective good solvent toluene were prepared.Oleylamine-stabilzed CdSe quantum dots with a peak emission at 525 nm and a particle diameter of 2.8 nm in hexane solution were purchased from STREM Chemicals Inc.The quantum dots were chosen since they show no absorption at the wavelength of the laser beam (l = 632.8nm).Through addition of twice the amount of ethanol absolute to the particle solution the quantum dots precipitated.Subsequently, the mixture was centrifuged at 11 000 rpm for 10 min.The hexane/ethanol mixture was decanted and the particles were dried under high vacuum for 2 hours.Afterwards, the quantum dots were dissolved in toluene which was then given to the block copolymer to attain concentrated solutions with a polymer concentration of 33.2 wt% and nanoparticle loadings of 0 wt%, 1 wt%, 2 wt%, and 3 wt% with respect to the amount of polymer. Birefringence measurement system and sample cell Time-and temperature-resolved in situ birefringence measurements are conducted with an Exicor 150AT from Hinds Instruments which measures magnitude (d) and fast axis orientation of the sample's optical retardation with high accuracy (0.01 nm) and time resolution allowing the acquisition of up to three data points a second.Owing to its special design moving parts in the optical train can be avoided and the alteration of measurement angles is not necessary.The HeNe laser beam is first polarized by a 451 polarizer and subsequently passes through a photo elastic modulator (PEM).After traversing through the sample the beam is divided in two parts by a beam splitting mirror.Both beams afterwards pass through an analyzer, an optical filter, and a photodetector.A lock-in amplifier processes the electronic signals which are further converted using a software algorithm to determine the magnitude of retardation and the angle of fast axis orientation.The sample cell used for the measurements is presented in Fig. 1.The electric field was applied in a home-built capacitor in which the block copolymer solution was sandwiched between two gold electrodes D of a length of 1.5 cm with an electrode spacing of 2 mm and a path length of light of 5 mm.The laser beam with a diameter of 1 mm and a wavelength of 632.8 nm propagated through the solution directly between the gold electrodes perpendicular to the electric field vector.Measurements were performed in the middle of the capacitor to avoid any influence by fringe fields close to the electrode edges.Above and below the capacitor, two glass slides I prevented solvent evaporation which were tightly screwed to the sample cell.For SAXS measurements the glass slides can be exchanged by kapton sheets covered with a thin layer of teflon R. Teflon pieces with the same dimensions as the glass slides (thickness: 1 mm; diameter: 3 cm) and a 1 mm wide centered hole directly above the middle of the capacitor, are pressed to the kapton sheets and screwed to the sample cell.This on the one hand fixes the kapton and on the other hand prevents voltage breakthrough to the outside of the cell.The temperature of the sample was precisely controlled using a circular Peltier element Q which was placed directly underneath the capacitor.Additionally, the temperature was monitored in solution via a PT-100 thermosensor C covered with thin glass.To prevent solvent degradation of the sample cell, this was built from polyether ether ketone (PEEK). Determination of the reorientational time constants via birefringence The sample was heated to 10 1C above the T ODT and subsequently cooled to 10 1C below the T ODT at a heating and cooling rate of 1 1C min À1 .While the lamella phase is characterized by a strong degree of form birefringence, the disordered phase is optically isotropic.Therefore, the T ODT can be determined by monitoring the decrease in phase retardation as a function of temperature.Afterwards, the block copolymer solution was quickly adjusted to the desired temperature.After cooling, the sample exhibits a microphase separated, isotropic multidomain structure with no preferred direction of orientation.Upon application of an electric field of 1 kV mm À1 , the increase in phase retardation is measured as a function of time.The data was fitted by a single exponential function to retrieve the overall reorientational time constant t. SAXS analysis of block copolymer reorientation mechanisms under application of an electric field While birefringence measurements offer excellent time resolution to analyze the reorientation kinetics of uniaxial block copolymer phases, an additional measurement technique is needed to draw conclusions about the underlying mechanism of reorientation.Publications by our group focused on unveiling the reorientation mechanisms of solutions of polystyrene-block-polyisoprene in toluene when exposed to electric fields. 23,25Since we use the same system for our measurements these findings can be exploited to interpret the birefringence measurements described in this publication.Hence, the most important findings are summarized below.As stated above electric field induced reorientation of block copolymer samples can proceed via three different mechanisms: Rotation of Grains (RG), Nucleation and Growth (NG), and Selective Disordering (SD). Previously, the reorientation mechanisms were distinguished by analysis of the time-evolution of the azimuthal intensity distribution of small angle X-ray scattering (SAXS) images upon inception of the electric field.A precise description of the measurements relevant to our study performed by Bo ¨ker et al. can be found in the ESI † (see Fig. S1). 23An important result of the measurements mentioned above is that the exploited reorientation mechanism is determined by the temperature at which the electric-field-induced alignment takes place with respect to the order-disorder transition temperature (T ODT ). Well below the T ODT , at lower temperatures, RG is the prevailing mechanism. 23In this case, the orientation of entire grains rotates as schematically demonstrated in Fig. 2(a).In the strong segregation limit (SSL), the average dimension of one region of coherence is larger and the formation of grain boundaries is thermodynamically unfavorable. 15Therefore the reorientation does not proceed via NG in this regime. Nucleation and Growth (Fig. 2(b)) proceeds via the formation of nuclei of lamella with an enthalpically preferred orientation parallel to the electric field vector and their subsequent growth.The starting point for the nucleation are defects, such as, for example, dislocations.NG is the prevailing mechanism at elevated temperatures, in the vicinity of the order-disorder transition, where the segregation between the blocks is weaker. 23In this weak segregation limit (WSL), the energetic penalty induced by the creation of boundary interfaces is lower whereby small nuclei which have a large boundary area in comparison to their volume are more readily formed.Furthermore, this region is characterized by a higher defect density and mobility which also favors NG. Compared to RG the reorientation through NG proceeds relatively slow whereby the time constants of reorientation at a given temperature can differ by an order of magnitude. 37nother important factor which greatly influences the reorientation of lamella domains under application of an electric field is the initial degree of order in the block copolymer sample.In misaligned samples, RG is the prevailing mechanism, while highly ordered block copolymers with lamella interfaces perpendicular to the electric field vector preferentially reorient via NG. 37he boundary between the two mechanisms is not sharp; at medium conditions both can coexist.Therefore, conditions prior to realignment are crucial. Analyzing block copolymer reorientation kinetics via birefringence While all block copolymer phases posses intrinsic birefringence, form birefringence is limited to uniaxial block copolymer phases.Intrinsic birefringence results from the orientation of block end-to-end vectors, and the stretching of block copolymer chains away from the block-block interfaces.Lodge and Fredrickson predicted that the intrinsic birefringence is directly proportional to the stretching of the chains defined by where L is the layer thickness and hh 2 i 0 is the mean-square unperturbed end-to-end distance. 39In strongly segregated lamella and cylindrical block copolymer systems (wN c 10) this represents a significant contribution to the overall birefringence strength of the sample and can even be comparable to or greater than the form birefringence. 39 The interfaces separating the microdomains are very narrow since the system wants to minimize the unfavorable A-B interfacial area.This results in perturbed, extended chain configurations and therefore in an entropic penalty.In our experiments the intrinsic birefringence of the block copolymer only has a minor contribution to the overall birefringence strength of the sample.The dilution of the polymer with toluene significantly lowers the T ODT to 62.5 1C, hence the individual polymer chains are largely unperturbed.This assures sufficient mobility for the reorientation process. Fredrickson and Leibler demonstrated that the theoretical descriptions of block copolymers in the molten state can likewise be applied to block copolymers in nonselective good solvents. 40In the following we will discuss how birefringence measurements can be utilized to analyze the reorientation kinetics of block copolymers as demonstrated by Amundson and coworkers. 41,42he polymer used in our study is characterized by a lamella microdomain structure.SAXS data of the exact same polymer in toluene solution can be found in previous publications by our group. 25Ruppel et al. determined an order-disorder concentration of 28.5 wt% for the polymer (S 42 I 58 108 ) in toluene solution at room temperature which is well below the concentration of 33.2 wt% used in our measurements.Fig. 3(a) displays a simplified illustration of a lamella stack, it's optic axis and the direction of the fast and the slow axis.A lamella microdomain pattern is characterized by a uniaxial optical anisotropy resulting in form birefringence.The axis of symmetry, also known as the optic axis, which is indicated by the green arrow in Fig. 3(a), is perpendicular to the lamella interfaces.Components of light propagating through the sample are exposed to different refractive indices n o and n e depending on whether their direction of polarization is perpendicular (ordinary wave (o), direction indicated by blue arrows) or parallel (extraordinary wave (e), direction indicated by red arrows) to the optic axis.In the case of negative uniaxial birefringence, as found in a lamella assembly, the fast axis coincides with the optic axis while the slow axis is perpendicular to it.Light polarized along the slow axis will be refracted with a refractive index n o while light polarized along the fast axis is exposed to a refractive index n e .Therefore, the ordinary wave will experience a phase retardation d with respect to the extraordinary wave. 43The magnitude of the resulting phase retardation d amounts to eqn (1). here l is the optical path length, l describes the wavelength of light while passing through the medium, C denotes the periodic lamella pattern, and y is the angle between the propagation direction of light and the unit wave vector of the lamella pattern e ˆk.Dn = n e À n o is the birefringence of the sample.The overall birefringence strength of a block copolymer is dependent on its degree of phase separation.In the strong segregation limit (SSL) the interfaces are sharp and a maximum form birefringence strength Dn 0 is attained. 41lock copolymer samples comprise several grains.When no orienting stimuli are applied to the sample these grains are randomly oriented.In Fig. 4(a) the evolution of the phase retardation signal (d) measured by the Exicor 150AT upon exposing a sample without quantum dots to an electric field of 1 kV mm À1 is plotted against time in seconds at three different temperatures.At the beginning of the measurement a phase retardation of around 0-10 nm is observed. Prior to application of the electric field, the block copolymer sample was heated to 10 1C above the order-disorder transition temperature (T ODT = 62.5 1C) at which the lamella microstructure disintegrates and approaches the disordered phase.Subsequently, the samples were cooled to 10 1C below T ODT at a cooling rate of 1 1C min À1 .By heating the sample above the T ODT and subsequent cooling into the phase separated state, a macroscopically isotropic multidomain structure is generated with no preferred orientation as schematically demonstrated in Fig. 4(b).This ensures that the initial conditions of alignment are comparable at all temperatures and nanoparticle loadings.When light propagates through the sample, each region of coherence induces a small phase retardation depending on its size, degree of phase segregation, and orientation.Amundson et al. 41 described the propagation of light through a macroscopically disoriented sample as a random walk on the poincare śphere -a spherical surface of unit diameter on which each point corresponds to a different polarization state of light. 41A schematic image of the poincare ´sphere is given in Fig. 3(b). Upon passing through the sample each encounter of light with a grain can be described as a small change of position on the poincare ´sphere, whereby direction and magnitude of the step are determined by the orientation, the birefringence strength and the size of the grain.This results in a series of uncorrelated small phase retardations.The associated trajectory on the poincare ´sphere is exemplified in yellow in Fig. 3(b).The overall birefringent phase retardation is determined by the position of the end point. The laser beam utilized in this study has a diameter of 1 mm while typical grain dimensions lie between 1 mm and 10 mm.Therefore, one ray of light traverses through various microstructural arrangements.The overall polarization state of the laser beam exiting the sample will be a mixed state composed of the individual end points of the random walks.Since these individual phase retardations are uncorrelated, in case of a macroscopically disordered sample the resulting overall d is of low value. After an electric field of 1 kV mm À1 was applied, the electric field-induced reorientation process set in and an increase in d over time was observed.The degree of alignment of a block copolymer sample is described by the orientational order parameter P 2 . A macroscopically disordered sample is characterized by P 2 = 0 (Fig. 4(b)) while a maximum alignment of lamella interfaces parallel to the electric field vector corresponds to an order parameter of À1/2 (Fig. 4(c)).For perfect alignment perpendicular to the electric field vector P 2 = 1 is found (Fig. 4(d)).The effective overall birefringence strength of the block copolymer sample composed of various small grains is directly proportional to the order parameter and amounts eqn (3). Therefore, the phase retardation can be used as a direct measure to analyze the reorientation kinetics.To retrieve the overall reorientational time constant t the data was fitted by a single exponential function d = d N + Ae (Àt/t) .As can be observed in Fig. 4 This journal is © The Royal Society of Chemistry 2016 55 1C a decrease in d N is observed.This can be attributed to the less defined interfacial boundaries between the block copolymer constituents in the weak segregation limit (WSL).Hence, the maximum attainable form birefringence strength Dn 0 and therefore also d N decreases upon approaching the order-disorder transition temperature.Since the interfaces are less defined the dielectric contrast between the blocks, which is the main driving force for electric field induced alignment, likewise decreases with rising temperature.Hence, it is expected that also the absolute value of the maximum attainable P 2 is lowered.The two factors are interlinked.Therefore, we cannot distinguish to which percentage a decrease in d N is induced by a reduction of P 2 or Dn 0 .Between 55 1C and 60 1C no further decrease in d N is observed, values fluctuate around 100 nm. Analyzing the switch in reorientation mechanism via birefringence The reorientation kinetics of the block copolymer sample without nanoparticles and with a nanoparticle loading of 1 wt% were analyzed via birefringence measurements at a field strength of 1 kV mm À1 in a temperature range between 30 1C and 60 1C.Owing to the intensive synchrotron SAXS studies on solutions of SI in toluene by Bo ¨ker et al. 23 described above direct conclusions on the reorientation mechanisms can be drawn from the birefringence measurements.Furthermore, a detailed synchrotron SAXS study on the alignment mechanisms of S 42 I 58 108 in toluene solution, which was also used for the measurements reported here, is described in a recent publication by our group and will be compared to our results in the course of the discussion. 25n the following, a description on how the switch in reorientation mechanism can be detected by birefringence measurements is given. In Fig. 5 the determined reorientational time constants t are plotted against the temperature for a sample with 0 wt% and 1 wt% CdSe nanoparticles.The reorientation kinetics are strongly dependent on the temperature of the sample. At first we will focus our discussion on the development of the reorientational time constants with increasing temperature of the sample with 0 wt% nanoparticles (black squares Fig. 5).Between 28 1C and 50 1C t decreases with rising temperature.In this low temperature regime all block copolymer domains reorient via RG. 23,25Upon raising the temperature, chain mobility is enhanced while the viscosity of the sample is lowered, leading to a faster reorientation.As long as all grains realign via RG, a decrease in t proportional to the increase in temperature is found as expected.The fact that the reorientational constant increases again upon further temperature elevation between 50 1C and 60 1C is most likely due to the onset of a slower realignment mechanism.According to previously published SAXS data NG sets in as an additional alignment mechanism in vicinity of the T ODT . 23,25 solution of the same polymer, with the same initial prealignment, reoriented at the same electric field strength of 1 kV mm À1 the results can directly be compared to our birefringence data.Fig. 5 shows that t reaches a maximum value at 60 1C, which is 2.5 K below T ODT .As described above the polymer reorients via a pure nucleation and growth mechanism at this temperature. 25In a temperature range between 50 1C and 60 1C an increase in t is observed with temperature.This can be explained by a coexistence of both mechanisms: with increasing temperature, the slower NG mechanism is increasingly preferred over RG.Since the percentage of grains realigned via NG increases with temperature, t rises upon temperature elevation.Unfortunately, it is not possible to resolve the individual time constants of the two reorientation mechanisms since the two reorientation processes are interlinked. 37ne might also interpret the slowing down of realignment kinetics with temperature between 50 1C and 60 1C as an effect of a simple reduction of dielectric contrast due to blurring of the boundary between the blocks in proximity of T ODT . 13This would lead to a reduced driving force for the reorientation process and hence to a worse overall alignment.To solve this issue we refer to Fig. S2 of the ESI, † where the maximum attainable birefringence retardation after realignment (d N ) is plotted vs. T.As stated above (d N ) is a measure of the order parameter after realignment and the degree of phase separation of the block copolymer.While (d N ) clearly decreases between 40 1C and 50 1C, no further decrease is observed in the temperature range in question.Hence, if the reduction of dielectric contrast should be reflected in a slowing down of realignment kinetics we would rather expect an increase in t between 40 1C and 50 1C where according to Fig. 5 a clear decrease is observed.In the temperature range where an increase of t is found, d N shows no clear trend with temperature and fluctuates around values of about 100 nm.Considering these observations and the information from the SAXS studies it is more probable that the slower alignment kinetics with rising temperature above 50 1C are induced by the onset of NG as an additional realignment mechanism.Upon comparison of the measurements with (yellow squares Fig. 5) and without particles (black squares Fig. 5), it becomes apparent that a low nanoparticle amount (1 wt%) is already sufficient to lower the temperature at which NG sets in as an additional mechanism by 10 1C down to a temperature of 40 1C.The results presented here are the first experimental evidence that the reorientation behavior is altered through the confinement of quantum dots. Two main factors are responsible for the onset of NG at lower temperatures upon addition of nanoparticles.On the one hand the incorporation of nanoparticles into block copolymers lowers the order-disorder transition temperature (T ODT ). 44T ODT of the samples was plotted against the quantum dot concentration (Fig. 6).At a nanoparticle loading of 1 wt% the T ODT is lowered by 4.5 1C compared to the sample without particles.On the other hand nanoparticles increase the defect density in the block copolymer compared to the pure samples.The quantum dots serve as nuclei for their preferential phase and support the formation and coarsening of grains.The activation energy for the nucleation and growth process is decreased and NG becomes the thermodynamically preferred mechanism of realignment. 45 Conclusion In conclusion, in situ birefringence measurements were applied to analyze the influence of quantum dots on the reorientation behavior of block copolymers upon inception of an electric field of 1 kV mm À1 .The birefringence measurements have the advantage of being inexpensive and easily accessible while at the same time offering a high time resolution to monitor the reorientation processes.The data presented here gives the first experimental evidence that the incorporation of nanoparticles into block copolymers leads to a preference of NG over RG. C. C. K. thanks the Fonds der Chemischen Industrie for financial support.The authors thank Werner Heckler for the introduction into the birefringence measurements, Guido Kirf for the construction of the sample cell, and Bernd Huppertz and Christoph Heeren for programming the temperature regulation. Fig. 1 Fig.1 (a and b) Photos of the sample cell used for the bulk analysis of block copolymers under the application of electric field.The block copolymer solution is sandwiched between two polished gold electrodes D. Two glass slides I tightly seal the chamber above and below the parallel plate capacitor preventing solvent evaporation as demonstrated in the schematic cross section image (d).Heating-and cooling rates can precisely be regulated via a circular Peltier element Q.A metal plate F below the Peltier element ensures rapid heat transfer to the surrounding.The Head B which seals the sample cell comprises a tiny PT100 thermosensor covered by thin glass C for precise in situ temperature determination.The high voltage power source is connected to the gold electrodes via E. (c) Engineering drawing of the sample cell. Fig. 2 Fig. 2 Schematic image of possible reorientation mechanisms of block copolymers upon inception of electric fields: (a) Rotation of Grains (RG) and (b) Nucleation and Growth (NG). Fig. 3 Fig.3(a) Simplified illustration of a single lamella stack, its optic axis and the direction of the slow and the fast axis with respect to the lamella interfaces.(b) Poincare ´sphere.The change in polarization state of light of a monochromatic laser beam passing through a block copolymer sample can be described as a random walk on the poincare ´sphere (here in yellow). (a) the maximum d values (d N ) differ from each other after application of the electric field.In Fig. S2 of the ESI † the d N values obtained from the fit functions are plotted against the alignment temperature.Between 40 1C and Fig. 4 Fig. 4 (a) Evolution of phase retardation of a sample of poly(styrene)block-poly(isoprene) (33.2 wt%) in toluene upon application of an electric field of 1 kV mm À1 at different temperatures.The wavelength of the laser beam used for the measurements amounts 632.8 nm.The phase retardation d increases proportional to the order parameter of the sample.The data was fitted by a single exponential function (black line).(b and c) Schematic illustration of the lamella microstructure before (b) and after (c) electric field-induced alignment.(d) Schematic illustration of a lamella microstructure oriented perpendicular to the electric field vector.P 2 is the corresponding order parameter. Fig. 5 Fig. 5 Time constants of the electric field-induced reorientation at 1 kV mm À1 for a sample of poly(styrene)-block-poly(isoprene) (33.2 wt%) in toluene without nanoparticles (black squares) and with 1 wt% of CdSe quantum dots (orange squares) as a function of temperature. Ruppel et al. analyzed a concentrated solution of S 42 I 58 108 in toluene, the polymer also used in this study, at 2.6 K below T ODT .When an electric field of 1 kV mm À1 was applied to the sample the polymer was shown to reorient via a pure nucleation and growth mechanism.Since Ruppel et al. used a toluene	2018-04-03T04:15:51.920Z	2016-10-12T00:00:00.000	{ "year": 2016, "sha1": "8bcebfdc5ce097794ccbd4627d115303589c5a2b", "oa_license": "CCBY", "oa_url": "https://pubs.rsc.org/en/content/articlepdf/2016/sm/c6sm01073c", "oa_status": "HYBRID", "pdf_src": "ScienceParseMerged", "pdf_hash": "8bcebfdc5ce097794ccbd4627d115303589c5a2b", "s2fieldsofstudy": [ "Materials Science", "Physics" ], "extfieldsofstudy": [ "Materials Science", "Medicine" ] }
16219874	pes2o/s2orc	v3-fos-license	Involucrin synthesis is correlated with cell size in human epidermal cultures. Late in terminal differentiation, human epidermal keratinocytes form an insoluble protein envelope on the cytoplasmic side of the plasma membrane. Involucrin, a soluble protein precursor of the envelope, is synthesized at an earlier stage of differentiation, both in the natural epithelium and in cultured keratinocytes. Because keratinocytes are known to enlarge during differentiation, we looked for a correlation between involucrin synthesis and cell size, using antiserum raised against the purified protein. We found that virtually no cultured epidermal keratinocytes with a diameter less than or equal to 14 micrometer contained involucrin, but most cells greater than 17 micrometer did. Using density gradient centrifugation, we were able to isolate a population of small cells containing almost no involucrin, as judged by immunodiffusion, PAGE, and immunoprecipitation. Large cells possessed translatable mRNA for involucrin, whereas small cells did not. We conclude that when cultured keratinocytes reach a certain size (approximately 14 micrometer in diameter) the specific mRNA for involucrin begins to accumulate and synthesis of the protein begins. Human keratinocytes (strain N, third to sixth passage), derived from newborn foreskin, were grown in the presence of lethally irradiated 3T3 cells (15) . In early experiments the medium was supplemented with 2096 fetal calf serum (15) . However, using the data of Barnes and Sato (2), S . Banks-Schlegel and H. Green (unpublished observations) found that in the presence of transferrin (5 ug/ml), insulin (5 pg/ml), and thiodothyronine (2 x 10-1`M), the concentration of fetal calf serum required for optimal cell growth could be reduced to 10%, and these conditions were used for later experiments. The keratinocytes grew equally well in either medium and the experimental results were the same. The medium also contained hydrocortisone at 0.4 lug/ml (15), and 10 -1°M cholera toxin (7) . Epidermal growth factor (EGF) prepared by the method of Savage and Cohen (21) was added to 5-10 ng/ml of medium beginning 2-3 d after subculture (16) . The medium was usually changed every 2-3 d and the day before cells were harvested. Before keratinocytes were harvested, any remaining 3T3 cells were removed by vigorous pipetting with an isotonic solution of EDTA (24) . The keratinocytes were then dislodged with a mixture of trypsin and EDTA. Cultures were usually harvested when three-quarters confluent . Labeling of Cell Protein with [ "S]Methionine The medium of keratinocyte cultures was replaced with medium containing a methionine concentration reduced from 30 mg/L to 3 mg/L. Approximately 100 yCi of I'Slmethionine (Amersham Corp., Arlington Heights, Ill . ; specific THE JOURNAL OF CELL BIOLOGY " VOLUME 90 SEPTEMBER 1981 738-742 ©The Rockefeller University Press " 0021-9525/81/09/0738/05 $1 .00 activity of 600-1300 Ci/mmol) in 8 ml of medium were added per 100-mm dish containing --4 x 10 6 cells. The cells were harvested after incubation at 37°C for 24 h. Preparation of Cell Extracts Harvested cells were washed once and resuspended in isotonic phosphate buffer containing 10 mM EDTA. The cells were sonicated and insoluble material was removed by centrifugation for 2 min at 12,800 g in a Brinkmann Eppendorf Centrifuge 3200 (Brinkmann Instruments, Westbury, N. Y.) . Separation of Keratinocytes of Different Size Keratinocytes were separated by density gradient centrifugation according to the method of Pretlow et al. (14,24). After harvesting, cells were washed once in isotonic phosphate buffer and resuspended in the buffer at 4°C. 5-15 X 106 cells were then layered on a 90-ml linear gradient of 25-65% (vol/vol) Percoll (Pharmacia Fine Chemicals, Piscataway, N. J.) in isotonic phosphate buffer. The gradients were prepared in 100-ml polycarbonate tubes which had been made wettable by treatment with sulphochromic acid (25) . Gradients were centrifuged at 4°C in an International refrigerated centrifuge with a swinging bucket rotor. Each centrifugation was carried out for 20 min at 800 g, with 4 min slow manual acceleration and deceleration. 5-ml fractions were collected by upward displacement with 60% (wt/vol) sucrose using a specially designed tapping device (14) (Halpro, Inc., Rockville, Md .) . Fractions were diluted with buffer and the cells recovered by low speed centrifugation at 500 g for 5 min. Scoring of Involucrin-containing Cells by Immunofluorescence and Determination of Cell Size Cells suspended in medium containing serum were either allowed to attach to collagen-coated glass cover slips for 5 min or air-dried onto uncoated cover slips. The cells were fixed in 3.7% formaldehyde in isotonic phosphate buffer at room temperature for 8 min, in methanol (-20°C) for 4 min, and in acetone (-20°C) for 2 min. After a brief rinse in buffer, the cells were stained for Immunofluorescence, using the procedure of Weber et al. (26) . Involucrin antiserum (19) or preimmune serum was used at a dilution of 1:20 in isotonic phosphate buffer, and fluorescein-conjugated goat antiserum to rabbit IgG (Miles Laboratories Inc., Miles Research Products, Elkhart, Ind.) at a dilution of 1:16. Preparations were examined with a Zeiss photomicroscope III, and several randomly selected fields were photographed. Cell diameters were measured from photographs, using a Bausch and Lomb measuring magnifier. Isolation of Poly A + mRNA and Translation In Vitro After density gradient centrifugation, total RNA was extracted from pellets of fractionated keratinocytes, using the guanidine procedure of Strohman et al. (23), as modified by Fuchs and Green (5). Poly A' mRNA was then isolated by affinity chromatography with oligo(dT)-cellulose (Type 3; Collaborative Research Inc., Waltham, Mass.) (4). Rabbit reticulocyte lysates were prepared by a method based on that of Schimkeet al . (22) and treated with micrococcal nuclease (Worthington Biochemicals Corp., Freehold, N. J.) (13) before use. The mRNA was translated at a final concentration of 9-18 pg/ml, in a total volume of l1 ILL. The reaction mixture, containing 1 pM [s6 Slmethionine (Amersham), was incubated for 1 h at 30°C. Methylmercury hydroxide (Alfa Div., Ventron Corp ., Danvers, Mass.) was added to some mRNA Solutions, to a final concentration of 2.5 mM (12) . Control samples were diluted with an equal volume of water. After 3-5 min, translation was initiated by addition of reticulocyte lysate and reaction mixture. Immunoprecipitation of Involucrin Involucrin was immunoprecipitated either from cell extracts or from translation products, using a method based on that of Kessler (9). Briefly, undiluted antiserum was added to the sample to an optimal ratio (usually 2 fd/ 10 id of cell extract) containing 10' M phenylmethylsulfonyl fluoride (PMSF) . After incubation at 4'C overnight, 70 pL of a 10% (wt/vol) suspension of formalin-fixed protein A-rich Cowan I strain Staphylococcus aureus (The Enzyme Center, Inc., Boston, Mass.) were added. The S. aureus had been washed three times in a butler solution consisting of 50 mM of Tris, pH 7.4, 0.05% Nonidet P-40, 1 mg/ ml of ovalbumin, 150 mM of NaCl, and 5 mM of EDTA. After 30 min at 4°C, the antibody-antigen complex bound to S. aureus was centrifuged and washed four times in buffer. Finally, theprecipitate wasresuspended in gelsample buffer and prepared for electrophoresis . PAGE Samples were electrophoresed in 8.5% acrylamide slab gels according to the method of Laemmli (10) . Gels were stained in a solution containing 0.1% Coomassie Brilliant Blue R250,50% methanol, and 10%acetic acid,and destained in a solution of 5% methanol and 10% acetic acid. Gels were fluorographed on Kodak RP2X-Omat film using the procedure of Bonner and Laskey (3). For quantitative immunoprecipitation studies, the film was prefogged (11) . Cell Size and the Presence of Involucrin Epidermal cells were disaggregated, collected by centrifugation, and resuspended in a small amount of medium containing serum. The cells were allowed to attach to cover slips, then were fixed and examined for the presence of Involucrin by Immunofluorescence . Cells containing Involucrin fluoresced brightly (Fig . 1 a), whereas cells lacking it had the same level of background fluorescence as cells stained with preimmune serum (Fig. 1 b) . The size distribution of cells possessing or lacking detectable Involucrin is shown in Fig. 2. Virtually no cells with a diameter of <14 Am contained the protein, but most cells >_ 18 pin did. About 30% of the entire population contained Involucrin . Isolation of Keratinocytes Lacking Involucrin Keratinocytes of different sizes can be separated by density gradient centrifugation, because the small cells have a greater buoyant density than the large ones (24). This method was used to confirm the relation between cell size and the presence of involucrin and to obtain a population of cells substantially free of involucrin. Epidermal cultures were disaggregated and the cells were washed in isotonic phosphate buffer, resuspended in buffer, and applied to Percoll gradients . After centrifugation, fractions collected from the gradients were examined microscopically and those containing cells of similar size were pooled . Typically, the largest cells, which had a buoyant density of -1 .04 g/ml, were found in the top six fractions, and the smallest, most dense cells (1.07 g/ml) in the bottom third of the gradient . The large-cell fraction was appreciably contaminated with small cells (Fig. 3 a), but there were very few large cells in the small-cell fraction (Fig. 3 b). Thus, although only -40% of the large-cell fraction showed positive staining, >96% of the cells in the small-cell fraction did not contain involucrin. For cells of a given size in either fraction, the proportion containing involucrin was in agreement with that found in the unfractionated population shown in Fig . 2 . Although immunofluorescence gives a reliable indication of the presence of involucrin in cells, it provides little information about the abundance of the protein . A quantitative measure of the amount of involucrin in different cell extracts was obtained by immunodiffusion, gel electrophoresis, and immunoprecipitation techniques . When serial dilutions of cell extracts were studied by Ouchterlony immunodiffusion, it was found that involucrin could be detected at a 1 :16 or 1 :32 dilution of largecell extracts, but only faint precipitin bands formed against small-cell extracts, even when undiluted. Extracts of large and small cells were subjected to gel electrophoresis in the presence of SDS and stained with Coomassie Brilliant Blue. The molecular weight of involucrin has been estimated previously to be 92,000 daltons (19) but, under the conditions of electrophoresis used here, the mobility of involucrin was somewhat less than that of rabbit muscle phosphorylase b (Sigma Chemical Co., St. Louis, Mo.). The extract of large cells contained an abundant protein with the same mobility as purified involucrin, whereas the small-cell extract showed only a very faint band at the same position. Immunoprecipitated extracts of cells labeled with [ 35Slmethionine for 24 h before harvesting were subjected to gel electrophoresis and fluorographed . The results were similar to those obtained on stained gels. Preimmune serum did not precipitate involucrin from extracts of either large or small cells (Fig. 4, tracks 3 and 6). The large-cell extract gave a strong band corresponding to involucrin, and this protein was selectively enriched by immunoprecipitation (Fig. 4, tracks 1 and 2). Densitometry of these tracks indicated that involucrin comprised, on the average, 5 .5% of the soluble cell protein . Extracts of small cells produced no definite band corresponding to involucrin, either before or after immunoprecipitation (Fig. 4, tracks 4 and 5), and by densitometry we calculated that involucrin was <03% of the soluble protein . Because the small-cell fraction isolated by density gradient centrifugation contained up to 4% of large cells, the observed difference in involucrin content of at least 18-fold between the two cell populations suggests that the protein is totally absent from small cells. Control of Involucrin Synthesis Having established that a population of keratinocytes which had not yet begun to synthesize involucrin could be separated from cells which had, we investigated whether these small cells also lacked translatable mRNA for involucrin . Poly A+ mRNA was prepared from fractionated cells and translated in vitro using a rabbit reticulocyte lysate. The translated proteins were precipitated with antiserum to involucrin and subjected to electrophoresis . Fig . 5 shows clearly that there was a considerable amount of mRNA for involucrin in the large cells (tracks 1 and 2) but virtually none in the small cells (tracks 4 and 5). By densitometry, we calculated that any involucrin mRNA present in small cells must be <6% of that in large cells . This is in reasonable agreement with the amount expected from contaminating large cells in the small cell fractions (Fig. 2). Pretreatment of mRNA with methylmercury hydroxide (12) did not improve the efficiency of translation of involucrin mRNA, suggesting that secondary structure is not a factor limiting translation. It may be concluded that involucrin synthesis is controlled by the amount of specific mRNA available for translation . Fig. 5 also shows that there is no substantial difference in the size of involucrin synthesized in vivo and in vitro; this appears to rule out substantial post-translational alteration of the size of the protein. DISCUSSION We have found a correlation between cell size and involucrin synthesis in cultured human keratinocytes . Using density gradient centrifugation, we have separated a population of small cells lacking involucrin from larger, less dense cells containing the protein . By preparing mRNA from the two cell fractions and translating it in vitro, we have shown that involucrin FIGURE 4 Electrophoretic analysis of involucrin from large and small cells. Extracts were prepared from large-and small-cell fractions after labeling with ["S]methionine for 24 h. After electrophoresis the gels were fluorographed. Extract of large cells (track 1) after precipitation with antiserum to involucrin (track 2) or with preimmune serum (track 3) . Small-cell extract (track 4), after precipitation with anti-serum to involucrin (track 5) or with preimmune serum (track 6) . Arrow indicates position of involucrin . synthesis is controlled by the cellular content of specific mRNA . The timing of some of the changes known to occur during Onset of cross-linking (granular layer of epidermis; upper layers of cultured Largest epithelium) FIGURE 6 Sequence of biosynthetic changes during terminal differentiation of keratinocytes. terminal differentiation of keratinocytes is summarized in Fig. 6. As previously described, the maturation of keratinocytes in epidermis is associated with a switch in synthesis from small to large keratins (6). It is clear that the onset of involucrin synthesis is controlled independently of large keratin synthesis: under the culture conditions used here, the switch to large keratin synthesis does not take place (6), whereas involucrin synthesis probably begins even earlier (with respect to the position of the cell) than in intact epidermis (1). In considering what signals lead to the appearance of mRNA for involucrin, it must be noted that the size of a cell is correlated with its position in the stratified epithelium and perhaps with the length of time since it ceased to divide. To analyze this problem further, it is necessary to have a means of evaluating the relation of cell size to biosynthetic properties independent of other variables. We thank Drs. E. Fuchs and B. Spiegelman for interesting discussions and helpful advice . These investigations were aided by grants from The National Cancer Institute.	2014-10-01T00:00:00.000Z	1981-09-01T00:00:00.000	{ "year": 1981, "sha1": "c0aa38a6ea5ef89a0325ccaffeb5246c9c84b592", "oa_license": "CCBYNCSA", "oa_url": "https://rupress.org/jcb/article-pdf/90/3/738/1075179/738.pdf", "oa_status": "BRONZE", "pdf_src": "PubMedCentral", "pdf_hash": "c0aa38a6ea5ef89a0325ccaffeb5246c9c84b592", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
16329586	pes2o/s2orc	v3-fos-license	Remote Sensing Integration of Concentration-area Fractal Modeling and Spectral Angle Mapper for Ferric Iron Alteration Mapping and Uranium Exploration in the Xiemisitan Area, Nw China The high-grade uranium deposits in the Xiemisitan area, northwestern China, are genetically associated with the faulting of felsic volcanic or sub-volcanic rocks. Ferric iron alteration indicates that oxidizing hydrothermal fluids percolated through the rocks. In this study, we measured the gamma-ray intensities of rocks in the Xiemisitan area and we propose a hybrid method for the mapping of ferric iron alteration using concentration-area fractal modeling and spectral angle mapper. The method enables ferric iron alteration to be distinguished from potash-feldspar granitic rocks. The mapping results were integrated with structural data to assist with exploration for uranium in the study area. Using this approach, six prospective areas of mineralization were proposed. Of these areas, two anomalies with high gamma-ray intensities of 104 and 650 Uγ were identified and verified by field inspection. These observations suggest that Enhanced Thematic Mapper Plus images are a valuable tool that can improve the efficiency of uranium exploration. Introduction Uranium is an important natural resource that is used in civilian and military industries.It is an incompatible element that becomes enriched in magma during the course of magmatic differentiation [1,2].Hence, felsic magma usually has a high uranium content and crystallized felsic igneous rocks can host uranium mineralization [3].The solubility of uranium depends on its valence (hexavalent uranium is more soluble in water-rich fluids than tetravalent uranium) [4,5], which is controlled mainly by the redox state of the geological environment [6].The redox state of terrestrial igneous rocks is usually close to the fayalite-magnetite-quartz (FMQ) [7,8] buffer.Under these conditions, uranium in magma is not oxidized from the tetravalent to the hexavalent state [6], thus preventing uranium from partitioning into a water-rich magmatic volatile phase that ultimately generates economic ore bodies [9][10][11].Fortunately, oxidizing surface waters percolating along geological faults or fractures can mix with post-magmatic hydrothermal solutions, resulting in a more oxidizing fluid that is capable of leaching uranium from volcanic or sub-volcanic rocks.Subsequently, precipitation from this fluid results in secondary uranium enrichment [12].Because the oxidation state of the fluid is higher than the hematite-magnetite (HM) buffer, minerals containing ferric iron (e.g., limonite and hematite) are formed during water-rock interaction.Ferric iron alteration is observed in many volcanogenic uranium deposits in northwestern China. Ferric iron alteration, felsic igneous rocks, and faulting are genetically related to uranium mineralization.Hence, these geological features are employed as criteria for uranium prospecting.Traditional exploration based on field survey is commonly inefficient and impractical because some regions, such as mountainous areas or the Gobi Desert in northwestern China, are inaccessible.In contrast, remote sensing technology employing low spatial and/or spectral resolution enables the rapid and low-cost collection of data from inaccessible regions, thus providing an alternative method to surveying for uranium resources. The Enhanced Thematic Mapper Plus (ETM+) is a satellite instrument that gathers multi-spectral remote sensing data free of charge, and these data are widely used in geological studies [13,14].Previous studies have proposed several methods and algorithms to identify alteration and lithology using the ETM+ [15][16][17][18][19].In particular, the Spectral Angle Mapper (SAM) is convenient and effective because the spectra used for identification is easily acquired from spectral libraries, Analytical Spectral Devices (ASD) spectrometer measurements, or from specified pixels in remote sensing images [20].However, the identification of hydrothermal alteration and lithology using SAM is strongly dependent on the selection of the maximum angle. The concentration-area (C-A) fractal model proposed by Cheng et al. (1994) [21] can be used to compute quantitative geochemical and geophysical anomaly thresholds, and it has been widely used in geological, geochemical, and geophysical exploration [22][23][24][25][26][27][28][29][30][31][32].Combining the C-A fractal model with the Crosta technique has been shown to be effective in identifying alteration zones in the Hashtjin area (Iran) [33], which suggests that the C-A fractal model is also useful for remote sensing studies.However, the Crosta technique may be sensitive to the presence of red potash-feldspar granitic rocks.Hence, this combined method may be unsuitable for the mapping of ferric iron alteration when such granitic rocks are present in the study area. In this paper, we propose a hybrid method for alteration mapping that uses the C-A fractal model to determine the maximum angle of SAM.We present a detailed description of data processing for the mapping of ferric iron alteration in the Xiemisitan area of northwestern China.The mapping result was confirmed by field inspection and was integrated with lithological and structural information to assist in exploration for uranium in the study area. Geological Setting The Xiemisitan area (46°45′-46°25′N; 84°50′-85°50′E), located west of the Junggar Basin in northwestern China (Figure 1a), contains the E-W running Xiemisitan mountain range at its center (Figure 1b).The mountain range is composed primarily of Devonian to Carboniferous volcanic and sub-volcanic rocks (including basalt, andesite, rhyolite, quartz porphyry, granite porphyry, and ignimbrite) and subordinate amounts of sedimentary rock (including carbonate, sandstone, and conglomerate) [34].Locally these rocks have been intruded by Hercynian magmatic rocks (including gabbro, diabase, diorite, monzogranite, and granite) and are cross-cut and bounded by the regional NE-trending Mengbulake Fault and the E-W-trending Bayinbulake Fault.Local faults and fractures are well-developed in the Xiemisitan area and are generally oriented NNE-SSW, NE-SW, and NW-SE. Geology of Baiyanghe Uranium Deposit The Baiyanghe uranium deposit in the Xiemisitan area (Figure 2) has a typical volcanogenic origin [35,36].The deposit is hosted by the Yangzhuang stock, which consists mainly of potash-feldspar granite porphyry and minor quartz porphyry.The weighted mean laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) U-Pb zircon age of the granite porphyry is 313.4 ± 2.3 Ma [34].The stock was emplaced into Devonian intermediate-acid volcanic and pyroclastic rocks, resulting in a friable and altered contact zone.The alteration minerals include hematite, limonite, fluorite, and chlorite.The deposit is located proximal to the E-W-trending Bayinbulake Fault (Figure 1b).The joints in the footwall of the fault can be divided into two groups.The first group is truncated by the second group (Figure 3a), suggesting that the direction of maximum stress changed in a manner that caused the fault style to change from reverse faulting to normal faulting.Some fractures (Figure 3b) in the porphyry stock are parallel to the Bayinbulake Fault and may have formed during the transition from compression to extension. Ore Minerals in Baiyanghe Uranium Deposit Ore minerals in the Baiyanghe uranium deposit include pitchblende and uranophane.Pitchblende is scattered and generally associated with fluorite veins lining small fractures or within the groundmass of felsic rocks, whereas uranophane mainly occurs on the surface of some felsic rocks with ferric iron alteration (Figure 3c).The U-Pb isotopic ages for the pitchblende can be divided into four stages, including 237.8-224 Ma, 197.8 Ma, 97.8Ma, and 30 Ma [36], suggesting post-magmatic and multi-stage mineralization of pitchblende.Because the REE patterns of pitchblende ore are similar to those of the granite porphyry, the granite porphyry may be the source of uranium in the Baiyanghe deposit that was further enriched through later hydrothermal activity [36]. Methodology As described in Section 1, SAM is a convenient and effective method for alteration mapping but its result is strongly dependent on maximum angle selection.The C-A fractal model, which permits anomaly thresholds to be quantitatively computed, provides an approach to determine the maximum angle of SAM.For a better understanding, the two methods are briefly described as follows. Spectral Angle Mapper The SAM algorithm calculates the spectral similarity between the image and reference spectra by treating them as vectors in n-dimensional space and computing the corresponding angle between them [32].The result is not affected by solar illumination factors because the angle between the two vectors is independent of vector length [32].In this study, we use the cosine value of the spectral angle to represent the degree of similarity between the image and reference spectra.High cosine values between the two spectra indicate high similarity, whereas low values indicate low similarity.The cosine value can be computed using Equation (1), where nb is the number of bands, ti is the reflectance of band i of the image spectrum, and ri is the reflectance of band i of the reference spectrum. (1) Concentration-Area Fractal Model The C-A fractal model proposed by Cheng et al. (1994) [21] can be used to separate geochemical and geophysical anomalies from background values in order to characterize the distribution of elements in a study area.The general form of the model is shown in Equation ( 2), where A(ρ) denotes the area with concentration values greater than the contour value ρ, v represents the threshold, and a1 and a2 are characteristic exponents.The area A(ρ) for a specified ρ is equal to the cell area multiplied by the number of cells with pixel values greater than ρ.The changes in slope between straight-line segments on a log-log plot and the corresponding values of ρ are used as cut-off (threshold) values to separate pixel values into components that represent different causal factors.In this study, we use the C-A fractal model to separate pixels with different cosine values and to determine the threshold value between different populations. Experiments This section presents a detailed description of data acquisition, processing, and modeling procedures for the mapping of ferric iron alteration in the study area. Reference Spectrum of Ferric Iron Ferric iron alteration is genetically associated with uranium mineralization and is therefore a valuable pathfinder for uranium exploration.Because of the low spatial resolution of ETM+ images and the possibility of mixed pixel effects, the spectrum of a ferric iron-bearing rock, rather than the spectra of pure ferric iron-bearing minerals, was measured in this study. The rock sample used for spectrum determination was collected from the Baiyanghe deposit.It is a granitic rock that contains uranophane and experienced hydrothermal alteration.Ferric iron bearing minerals can be observed on the surface of the sample.Some feldspars have altered to clay minerals.The measurement was carried out using an ASD Fieldspec Pro spectrometer at National Key Laboratory of Science and Technology on Remote Sensing Information and Image Analysis.The spectral range of the instrument is 350 to 2500 nm.The spectral resolution is 3 nm between 350 and 1000 nm, and 10 nm between 1000 and 2500 nm.The instrument was set up according to the manufacturer's instructions described in the Labspec4 user manual.An accessory light source was employed to measure spectra in lieu of sunlight.During the measurement, a relatively flat rock surface containing ferric iron alteration was chosen to avoid potential interference from background environmental fluorescence.A white reference light was introduced during the measurement procedure to optimize and calibrate the instrument.Five spectra of the rock sample were measured and recorded using the RS 3 software.The data were converted into ASCII file format using ViewSpecPro software (see Supplementary Table 1). The spectra listed in Supplementary Table 1 was converted to Spectral Library File using Spectral Library Builder function of ENVI 5.0 software, and then resampled to Landsat TM7 sensor spectra with Spectral Library Resampling function.The reflectance values of resampled spectra and their average values, i.e. the reflectance values of the reference spectrum (used to compute cosine values of spectral angles) are listed in Table 1. Processing of ETM+ Image The original ETM+ image of the study area was acquired during the dry season on 27 June 2000 by the Landsat 7 satellite (Path = 145, Row = 28) and processed to generate a standard product by the United States Geological Survey (USGS) on 22 February 2008.The USGS product was downloaded from the Earth Explorer platform (www.earthexplorer.usgs.gov).It contains one panchromatic band (band 8) and seven multi-spectral bands (bands 1-7).The spatial resolution of band 8 is 15 m and the resolutions of the multi-spectral bands are 30 m except for band 6 (60 m).The cloud content of the product is less than 1%.Band 6 was excluded from the image dataset because it is a thermal infrared band that may be influenced by temperature variations related to elevation changes.and will be used to compute cosine values of spectral angles.** r i is the reflectance of No. i band of reference spectrum(See Equation ( 1)). The ETM+ image was oriented to acquire part of the Xiemisitan area and was then calibrated using the "Landsat Calibration" function in ENVI 5.0 software.An atmosphere correction was carried out using the "Quick Atmosphere Correction" function.Since pixel values in the calibrated and corrected image were multiplied by a constant to retum integer values by ENVI software after atmosphere correction, the image was normalized using the "Band Math" function by dividing by 10,000. The normalized ETM+ image and the reference spectrum listed in Table 1 were used to compute the cosine values of spectral angles using the "Band Math" function following the method described in Section 3.1.The computation expression is provided in Supplementary 1 and the result is shown in Figure 4. C-A Fractal Modeling In this study, the cell area (i.e., one pixel) is 30 × 30 m and the total number of cells in the cosine image (Figure 4) is 5.031 × 10 6 .The contour value ρ denotes the pixel value in the cosine image, and it ranges from 0.6723 to 0.9993. The "Density Slice" function in ENVI 5.0 software was used to divide the cosine map into 3270 gray levels each of which displays a variation of ~0.0001 in the cosine value.The density slice image was saved as a "Class Image" for statistical analysis, using the "Class Statistic" function.The results (Supplementary Table 2) were exported to Microsoft Excel 2007 to create a histogram (Figure 5a) and log-log plots (Figure 5b). Results The mapping result of ferric iron alterations for the study area using the hybrid method of SAM and C-A fractal model is presented in this section.Additionally, we determined the gamma intensities of rocks in the Xiemisitan area for further uranium exploration application.5b were fitted using least square method. Ferric Iron Alteration Mapping Generally, the mapping results for iron oxides can be classified into the background, low intensity, and high intensity areas [33].We, therefore, identified three populations of data based on the log-log graph shown in Figure 5b.Three linear segments were fitted to the data using the least squares method.The cut-off (threshold) value was determined from the x-coordinate of the intercept between two adjoining line segments, at which the product of the linear correlation coefficients is a maximum.The calculated values for thresholds v1 and v2 (Figure 5b) are 0.9972 and 0.9904, respectively.These thresholds were used to divide the map of cosine values into a red area where pixel values are greater than 0.9972, and a yellow area where pixel values range from 0.9904 to 0.9972 (Figure 6). A belt of ferric iron alteration along the contact zone between the porphyry stock and the Devonian intermediate volcanic rocks (north of the Baiyanghe uranium deposit) offers a good opportunity to test the mapping results.The E-W trending belt is ~4 km long and 25-45 m wide, and consists mainly of potash-feldspar granite porphyry that contains ferric-iron-bearing minerals (e.g., limonite and hematite) and minerals indicative of argillic alteration.As shown in Figure 6, the mapping results for ferric iron alteration can be divided into two groups.The first group contains pixels with high cosine values (red pixels in Figure 6) that are similar to the reference spectrum of ferric iron alteration.This group is spatially consistent with the area where ferric iron minerals are abundant.In contrast, the second group contains pixels with low cosine values (yellow pixels in Figure 6) that are markedly different to the reference spectrum, consistent with their location in areas with minor or no ferric-iron-bearing minerals.The distribution of the second group of yellow pixels coincides with the distribution of potash-feldspar porphyry in some places (Figure 7, also see Supplementary 2).Collectively, these observations suggest that the combination of SAM and the C-A fractal model is not only useful for the mapping of ferric iron alteration, but also for distinguishing between ferric iron alteration and potash-feldspar porphyry rocks, which are both genetically related to uranium mineralization in the Baiyanghe deposit.Hence, the mapping results will be valuable for uranium exploration in the Xiemisitan area. Gamma Intensities of Rocks Measuring the gamma-ray intensities of rocks is an effective approach to identifying rocks that are enriched in uranium.In this study, 500 magmatic and sedimentary rock samples were collected from the Xiemisitan area.In addition, ferric iron-bearing rocks from the Baiyanghe uranium deposit were collected.The gamma-ray intensities of these rocks were measured using an HD-2000 gamma indicator developed by the Beijing Research Institute of Uranium Geology, Beijing, China.The intensity unit is Uγ.Please refer to Supplementary 3 or www.bjhdkj.comfor more information about the indicator.The gamma-ray intensities of felsic magmatic and volcanic rocks (35 to 42 Uγ) in the Xiemisitan area are higher than those of andesitic (22 to 29 Uγ) and mafic rocks (17 to 20 Uγ), which is expected because of the incompatible behavior of uranium during magmatic differentiation.Sedimentary rocks, including sandstone, conglomerate, and carbonate, usually have low gamma-ray intensities of 23 to 29 Uγ.The ferric iron-bearing rocks in the Baiyanghe deposit yield higher gamma-ray intensities (mean of 67 Uγ), and rocks containing uranophane yield even higher gamma-ray intensities (mean of 953 Uγ; Table 2).Previous studies reported uranium contents of 13.43 ppm, 101.09 ppm (~0.01%), and ~1265.58ppm (0.13%) for felsic magmatic rocks, ferric iron rocks with no uranophane, and ferric iron rocks with uranophane, respectively [40].Since the anomaly threshold of uranium is 0.01% (67 Uγ in this study) in China, we use 67 Uγ as the cut-off value to distinguish uranium anomaly from background. Discussion In this section, we first discuss the advantages of the hybrid method of SAM and C-A fractal model, then integrate mapping results with structural data to assist with exploration for uranium in the study area. Advantages of the Method The C-A fractal model can be used to identify different data populations for image display and to determine the threshold values that separate each population.It is useful not only for geochemical and geophysical studies but also for remote sensing because the model generates different populations of pixel values and takes into account the spatial and geometrical properties of real-world ground features [33].The SAM method enables rapid mapping and suppresses the influence of shading effects to accentuate the reflectance characteristics of the target [20].In addition, the reference spectrum used for SAM mapping is easy to obtain. Previous studies proposed that the C-A fractal model could be combined with the Crosta technique to map ferric iron alteration [31][32][33].However, this method may be compromised by the red color of potash-feldspar granitic rocks, which causes high reflectance values in band 3 and low values in band 1 that are similar to those of ferric-iron-bearing minerals.The method presented in this study, which combines SAM and the C-A fractal model, enables a more robust identification of ferric iron alteration because it uses a greater number of spectral characteristics.As shown in Figure 6, the method can distinguish ferric iron alteration from potash-feldspar granite porphyry.The method can potentially also be used for mapping other types of hydrothermal alteration using appropriate reference spectra.The whole data processing procedure can be completed using the ENVI 5.0 and Microsoft Excel software packages without additional programming. Application to Uranium Exploration Faults and fractures play an important role in fluid migration and uranium mineralization.These structures can be identified by processing a ETM+ image to clearly show changes in rocks units caused by displacement and deformation [41].In this study, structural separation was identified through the interpretation of ETM+ images using a band combination (Red = band 4; Green = band 3; Blue = band 1) and was based on evidence such as vegetation change, the deviation of stream paths, and abrupt lithology changes as depicted by changes in the tones and textures of images.The result is shown in Figure 8. The faults identified with this method and the maps of ferric iron alteration and potash-feldspar granite were used for uranium exploration in the Xiemisitan area.Six prospective areas of uranium mineralization were recognized and selected for measurement of gamma-ray intensity.These areas are denoted by the numbers in Figure 9. The radioactive element content of each area was inferred from the gamma-ray intensity measured using the HD-2000 gamma indicator.Two gamma-ray anomalies from area 5 were identified and had gamma-ray intensities much higher than those of felsic rocks (35 to 42 Uγ) in the Xiemisitan area (as described in Section 2.3).The first anomaly had a gamma-ray intensity of 104 Uγ and was found in fractures in a granite porphyry stock (Figure 10a) where ferric iron alteration and faults are well developed.The second anomaly was also found in granite porphyry affected by ferric iron alteration (Figure 10b) and is characterized by a gamma-ray intensity of 650 Uγ.The two newly discovered anomalies suggest that ETM+ images are valuable for uranium exploration.Although two anomalies were recognized in area 5, other areas show no apparent gamma-ray anomalies despite the occurrence of ferric iron alteration.This apparent discrepancy can be attributed to the surface oxidation of felsic rocks containing iron-bearing minerals.For example, ferric iron alteration is well developed on the surface of the granite porphyry in area 6, but does not occur within the fresh interior of the rock (Figure 11).In this case, the ferric iron alteration was related to surface oxidation rather than interaction with hydrothermal fluids.The gamma-ray intensities of the granite porphyry range from 35 to 47 Uγ, suggesting no uranium enrichment in this area. The surface oxidation of iron means that the efficiency of uranium exploration using ETM+ images alone is severely limited because false uranium prospects will be identified.Fortunately, uranium-bearing ferric iron alteration is usually characterized by higher gamma-ray intensity (sometimes by one order of magnitude) than uranium-barren rocks.Hence, airborne gamma-ray measurements are a useful tool for uranium prospecting.Since ETM+ images are freely available and easy to acquire and process, we propose a two-stage procedure for uranium exploration.The ETM+ image is used in the first stage to identify possible uranium-enriched areas based on regional lithology, ferric iron alteration mapping, and the presence of faults.In the second stage, these areas are further evaluated by airborne gamma-ray spectrometer or ground-based gamma-ray measurements to identify areas of high intensity. Conclusions The gamma-ray intensity of rocks is important for uranium exploration.The gamma-ray intensities for the felsic, intermediate, mafic, and sedimentary rocks in the Xiemisitan area are 35-42, 22-29, 17-20, and 23-29 Uγ, respectively.Rocks containing uranium-bearing ferric iron alteration have gamma intensities of about one order of magnitude (953 Uγ) higher than those of uranium-barren rocks.Hence, high gamma-ray intensities can be used as a pathfinder for uranium exploration in the Xiemisitan area. A previous method combined the C-A fractal model with the Crosta technique to map ferric iron alteration, but this method is compromised when potash-feldspar granitic rocks are present in the study area.In contrast, the hybrid method of SAM and the C-A fractal model presented in this study uses a greater number of spectral characteristics and thus can robustly distinguish ferric iron alteration from potash-feldspar granite porphyry.Because the reference spectrum is easy to obtain, the hybrid method can potentially be used to map other types of hydrothermal alteration using appropriate reference spectra.In addition, the processing procedure can be conveniently completed using the ENVI 5.0 and Microsoft Excel software packages without additional programming. The case study reported here demonstrates that ETM+ images can be used to identify ferric iron alteration, potash-feldspar granitic rocks, and geological faults.Hence, ETM+ images are valuable during the early stages of uranium exploration.However, it was also found that the surface oxidation of felsic rocks containing iron-bearing minerals might limit the efficiency of uranium prospecting.Therefore, validation of prospective areas of mineralization using airborne or ground-based gamma-ray measurements is essential for efficient uranium exploration using ETM+ images. Figure 1 . Figure 1.(a) Location of the Xiemisitan area in NW China.(b) Elevation map derived from ASTER illustrating the physiography of the Xiemisitan area. Figure 2 . Figure 2. (a) Geological map of the Baiyanghe Uranium deposit.(b) Geological profileshows that uranium ore bodies are distributed along the fracture zone between granite porphyry and the Taerbahata formation (modified after[36]). Figure 3 . Figure 3. (a) Two groups of joints, in which the second group is limited by the first group.(b) Penetrative fractures in the porphyry stock.(c) Uranium mineralization is associated with ferric iron alteration. Figure 4 . Figure 4. Map of cosine values of spectral angles between image and reference spectra.The image spectra were obtained from the pro-processed ETM+ image, and the reflectance values of reference spectrum are listed in Table1. Figure 5 . Figure 5. (a) Histogram graph of area versus cos(α) in the Xiemisitan area.(b) Log-log plot of A(ρ) versus cos(α) in the Xiemisitan area.The straight lines shown in Figure 5b were fitted using least square method. Figure 6 . Figure 6.Distribution of ferric iron bearing rocks mapped by combination method of SAM and C-A fractal model.Cells in red area have pixel values greater than 0.9972 and those in yellow area have pixel values in the range of 0.9904 to 0.9972. Figure 7 . Figure 7. Photograph of ferric iron alteration belts in the north of the Baiyanghe uranium deposit. Figure 9 . Figure 9. Distribution of uranium prospects (red pixels represent ferric iron, while yellow pixels represent felsic magmatic rocks). Figure 10 . Figure 10.(a & b) Two radioactive anomalies (Gamma intensities were measure by HD-2000 gamma indicator developed by Beijing Research Institute of Uranium Geology).High quality photographs can be found in Supplementary 4-7. Figure 11 . Figure 11.Ferric iron is well developed on the surface of the granite porphyry, but does not occur in the fresh part of the rock in area 6. Table 1 . Reflectance values of resampled spectra and their average values. *The average values were calculated from the reflectance of resampled spectra (RS1 to RS5) Table 2 . Gamma intensities for different rocks in the Xiemisitan area.	2016-04-23T08:45:58.166Z	2015-10-22T00:00:00.000	{ "year": 2015, "sha1": "e600ff21ac4387a6b607fb67727b38ed3ecc949b", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2072-4292/7/10/13878/pdf?version=1445509761", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "e600ff21ac4387a6b607fb67727b38ed3ecc949b", "s2fieldsofstudy": [ "Geology" ], "extfieldsofstudy": [ "Geology", "Computer Science" ] }
260833421	pes2o/s2orc	v3-fos-license	Autoimmunity against laminin 332 Laminin 332 is a heterotrimeric structural protein of the basal membrane zone (BMZ) of the skin and adjacent mucosal tissues. The importance of laminin 332 for the structural integrity of the BMZ is demonstrated by mutations in any of the three genes encoding for its three chains causing variants of junctional epidermolysis bullosa. Autoimmunity against laminin 332 is observed in mucous membrane pemphigoid (MMP) and in the rare patients with orf-induced pemphigoid. MMP is an autoimmune blistering disease with predominant mucosal manifestations and autoantibodies against the BMZ of the skin and orifice-close mucous membranes. The main autoantigens of MMP are type XVII collagen (BP180) and laminin 332 targeted in about 80% and 10-20% of patients, respectively. An increasing number of studies has highlighted the association of anti-laminin 332 MMP and malignancies that can be revealed in about a quarter of these patients. This data has led to the recommendation of current guidelines to assay for anti-laminin 332 reactivity in all MMP patients. The present review focuses on anti-laminin 332 MMP describing clinical features, its pathophysiology, and detection of serum anti-laminin 332 IgG. In addition, the available data about the occurrence of malignancies in anti-laminin 332 MMP, the underlying tumor entities, and its biology are detailed. Introduction Laminin 332 is a heterotrimer and essential structural protein of the basal membrane zone (BMZ) of the skin, adjacent mucosal tissues including the mouth, pharynx, larynx, trachea, esophagus but also kidney, lung, and small intestine (1). The importance of laminin 332 for the structural integrity of the BMZ is demonstrated by mutations in any of the three genes LAMA3, LAMB3 and LAMC2, that cause a variant of junctional epidermolysis bullosa (2,3). Autoimmunity against laminin 332 is observed in the autoimmune blistering disease mucous membrane pemphigoid (MMP) and in the very rare patients with orf-induced pemphigoid (4,5). Furthermore, autoantibodies against laminin 332 have been described in individual patients with bullous pemphigoid, anti-p200 pemphigoid, and epidermolysis bullosa acquisita in addition to the disease-typical autoantibodies against, i.e. BP180/type XVII collagen, p200 protein, and type VII collagen, respectively (6)(7)(8)(9)(10)(11). The present review focuses on anti-laminin 332 MMP summarizing clinical features, its pathophysiology, and detection of serum anti-laminin 332 IgG. In addition, the current data about the association between antilaminin 332 MMP and malignancies are highlighted. The current review is dedicated to the late Detlef Zillikens, director and chair of the Department of Dermatology, University of Lübeck, Germany. Detlef Zillikens has been one of the leading experts on autoimmune blistering diseases. With an enormous workload and his friendly, optimistic, supportive, and caring nature he has established in Lübeck one of the world largest research hubs for these disorders. As one of his first students in 1993, close collaborator, mentee, and friend, E.S. owes him the greatest thanks for constant support, motivation, and fruitful discussions. S.P. got to know Detlef Zillikens in 2016 when starting her PhD thesis and owes him the greatest respect and thanks for his support of a young scientist and incessantly enjoyment of research. He was able to close the gap between science and clinic due to his dedication for both disciplines and his view of the entire picture. Both authors will strive to continue Detlef's work and guard the best memories of him. Laminin 332 Laminins are cross-or T-shaped heterotrimers of an a, b and g chain with three short arms (single chains) and one long arm formed by all three chains (12). Laminins are integral proteins of the BMZ of the skin and surface-close mucosal tissues. Here, they are essential components of the anchoring filaments connecting the hemidesmosome with type VII collagen (13). Their physiological functions include adhesion of the epidermis to the dermis and epithelium to the lamina propria, respectively, cell migration and, cell signaling (12). Mucous membrane pemphigoid MMP is a clinically and immunopathologically heterogeneous disease defined as pemphigoid disorder with prevailing involvement of orifice-close mucosal tissues (18). As a pemphigoid disorder, MMP is characterized by autoantibodies that bind to the BMZ of the skin and/or mucosa (19,20). Clinical heterogenicity is reflected by the involvement of different mucosal sites, most frequently the mouth (in about three quarters of patients) and conjunctivae (in about 50-65% of patients) followed by nasopharynx and genitalia, and more rarely, larynx, esophagus, and trachea. In about a quarter of patients, in addition to mucosal manifestations, skin lesions are present ( Figure 1) (21, 22). The high disease burden of MMP is due to frequently painful oral and genital lesions, life-threatening complications such as airway obstruction and esophageal strictures, conjunctival disease leading to vision impairment and finally, blindness, and the association with a malignancy in about a quarter of patients with anti-laminin 332 reactivity (22). Immunopathological heterogenicity stems from the different target antigens and the autoantibody isotype. While in most MMP patients, autoantibodies belong predominantly to the IgG isotype, the majority of patients also reveal IgA autoantibodies, and in some, the autoantibody response is restricted to IgA (22-24). BP180 (type XVII collagen) as main target antigen in MMP is recognized by about 70-80% of patients followed by laminin 332 in 10-20% of patients. In less than 5% of MMP patients, type VII collagen is recognized. Reactivity against BP230, that can be found in 10-30% of cases, is nearly always accompanied by autoantibodies against one of the three other target antigens (21, 24). In some MMP patients, autoantibodies against a6b4 integrin have been described (25-28). The relevance of these a6b4 integrin-specific antibodies in MMP is, however disputed (24, 29). Patients with mostly mucosal manifestation and predominant IgA reactivity, that previously may have been classified as linear IgA disease, and those with autoantibodies against type VII collagen previously diagnosed as epidermolysis bullosa acquisita, are now regarded within the spectrum of MMP (21). Few data about the frequency of MMP are available. With an incidence between 1.3 and 2.0/million/year in France and Germany, respectively, and a prevalence of 24.6 patients/million in Germany, MMP is certainly a rare disease (30-33). MMP arises independently of ethnicity and geographical region, mainly affects individuals in the 7 th and 8 th decennium, and appears to be more frequent in females (22). Diagnosis of MMP, like in all autoimmune blistering diseases, is grounded on three pillars; clinical manifestations, direct immunofluorescence (IF) microscopy, and serology (20, 24). The clinical prerequisite is predominant mucosal involvement. Direct IF reveals linear deposits of IgG, IgA, and or C3 at the cutaneous or mucosal BMZ in a non-lesional biopsy ( Figure 2). Since the initial biopsy only provides a sensitivity of 50-70% depending on the biopsy site, current guidelines recommend to repeat the biopsy for direct IF at least once after an initially negative result (24, 29). Detecting of circulating autoantibodies against the abovementioned antigens is complex, mainly based on in-house assays, and reviewed elsewhere (22, 24). The detection of anti-laminin 332 IgG is detailed below. Treatment of MMP is greatly hampered by the lack of randomized controlled studies. National and international guidelines propose treatment regimens (24, 34-37). The S3 European guidelines included a systematic literature review and recommend dapsone, methotrexate, tetracycline, and topical corticosteroids as first line treatment for mild and moderate MMP. For severe MMP, dapsone plus cyclophosphamide and/or oral corticosteroids are suggested and, if not successful, dapsone plus rituximab followed by latter two drugs combined with highdose intravenous immunoglobulin (24). A slightly different stepladder approach was published in the recent German S2k guideline (29). Anti-laminin 332 mucous membrane pemphigoid In 1992, laminin 332 has been described as a target antigen in MMP by Kim Yancey and co-workers (4). Since then, numerous case reports and case series have reported IgG serum autoantibodies against this protein. It was only in 2019, when a highly standardized and specific assay for serum anti-laminin 332 IgG became widely available (38). Clinical appearance of anti-laminin 332 mucous membrane pemphigoid A patient with anti-laminin 332 MMP can clinically not be differentiated from a MMP patient with autoimmunity against BP180 or type VII collagen. In a systematic review of published cases and cohorts, Amber et al. reported significantly more pharyngo-laryngeal and oro-pharyngo-laryngeal involvement in MMP patients with reactivity against laminin 332 (39). In the so far largest study with 133 anti-laminin 332 MMP patients from Kurume, Japan, the oral cavity was the by far most frequently affected mucosal site (in 89% of patients) followed by conjunctivae (in 43%), pharynx (in 19%), larynx (15%), genital mucosa (in 11%), nasal mucosa (in 6%), and esophagus (in 3%) (40). Compared with MMP patients independent of the target antigen as recently reported in 154 MMP patient and as reviewed by Du et al., nasal lesions appear to occur less frequently in anti-laminin 332 MMP compared to 20-40% in all MMP patients, while oral lesions may be slightly more prominent (in 80-85% of all patients) (22, 41). These differences have, however, not been systematically evaluated and may also be related to the different ethnicity or other so far unrecognized factors. Recently, a significant association of laminin 332-reactive MMP with male sex was reported (41). The most striking and clinically relevant feature that differentiates anti-laminin 332 MMP from MMP with other autoantibody reactivities, the association with malignancies in about a quarter of patients, is detailed below. Detection of anti-laminin 332 reactivity Several methods have been applied to detect anti-laminin 332 reactivity in skin and mucosal biopsies as well as in serum. Direct and indirect immunogold electron microscopy show deposits of immunoreactants at the lamina lucida/lamina densa interface of the BMZ in anti-laminin 332 MMP. In patients with autoantibodies against BP180 or type VII collagen, immunoreactants label the lamina lucida or the subbasal lamina-anchoring fibril zone, respectively (4, 42-45). Direct immunogold electron microscopy requires, however, fresh biopsy material that needs to be processed within hours and is only performed in few centers worldwide (46). For the detection of serum autoantibodies against laminin 332, indirect immunogold electron microscopy is unpractical and as such, several in-house assays have been described including (i) immunoprecipitation of radiolabeled keratinocytes that was also FIGURE 2 Linear deposits of complement C3 at the basement membrane zone by direct immunofluorescence microscopy of a perilesional biopsy in a patient with mucous membrane pemphigoid. (53), (j) primary human oral mucosal keratinocytes (54), and (k) immortalized human oral mucosal keratinocytes (54), and (iii) ELISA. When immunoprecipitation was compared to immunoblotting with five different substrates, i.e. (b-f), immunoprecipitation was identified as the most sensitive method followed by Western blotting with extracellular matrix of cultured human keratinocytes (II f) (50). In sera with reactivity against the cutaneous BMZ by indirect IF microscopy on human skin, indirect IF on laminin 332-deficient skin from patients with junctional epidermolysis bullosa (being unreactive on latter substrate) as well as the fluorescence overlay antigen mapping on human salt-split skin are elegant methods to determine autoantibodies against laminin 332 (61). Another test based on indirect IF, the so-called footprint assay, demonstrated that anti-laminin 332 serum IgG can be detected in the extracellular matrix of cultured primary keratinocytes after removal of the cells from the glass coverslips. Here, the extracellular matrix of the removed individual keratinocytes appear as traces or "footprints" that can be visualized by anti-laminin 332 antibodies followed by FITC labelling (59). A breakthrough was achieved by Goletz et al. who described an indirect IF test based on the HEK293 cells that recombinantly express the laminin 332 trimer on their cell surface ( Figure 3). As negative control, HEK293 cells transfected with an empty vector are used. These cells are applied using the BIOCHIP ® mosaic technology, i.e. several substrates are placed together in a single incubation field of a laboratory slide (62-65). When in an international multicenter study, 93 anti-laminin 332 MMP patient sera and 315 sera from other autoimmune blistering diseases including 153 sera from anti-laminin 332 negative MMP patients, non-inflammatory dermatoses, and heathy blood donors were probed, a sensitivity of 84% and a specificity of 99.6% were observed (38). This assay has subsequently been validated by other groups (66,67). When the BIOCHIP ® technology-based assay has recently been compared with the footprint assay using 54 anti-laminin 332 MMP sera and together 50 sera from patients with pemphigus vulgaris and healthy blood donors, both assays revealed a specificity of 100% with a slightly higher sensitivity of the footprint assay (100% versus 96.3%) (60). When 35 sera of originally laminin 332-unreactive sera were subjected to both IF tests, 3 were reactive in the BIOCHIP ® assay and 7 in the footprint assay. These data show that the footprint test may be more sensitive, whereas the advantage of the BIOCHIP ® assay is its high standardization and wide availability (60). Reactivity against the different laminin chains varied considerably between studies. In 113 Japanese patients with antilaminin 332 MMP, the g2 chain was most frequently recognized (in 58% of patients) followed by a3 and b3 targeted in 49% and 36% of patients, respectively (40). In contrast, Goletz et al., using the BIOCHIP ® technology-based IF assay in an international multicenter study with 93 sera, reported IgG4 reactivities against the a3, b3, and g2 in 43%, 41%, and 13% of patients (38). These discrepancies maybe most likely due to the different study populations or detection methods. In individual MMP patients, IgA and IgE antibodies against laminin 332 have also been reported (68, 69). Since anti-laminin 332 MMP is associated with a malignancy in about a quarter of patients as detailed below, national and international guidelines recommend the detection of anti-laminin 332 serum IgG in all patients that show dermal binding by indirect IF on human salt-split skin or were unreactive in this assay (24, 29). A suggested diagnostic pathway for anti-laminin 332 MMP is depicted in Figure 4. Proposed diagnostic algorithm for mucous membrane pemphigoid (MMP). Adopted from (22, 24, 70). 1 in the oral cavity, a non-lesional biopsy is equally sensitive compared to a perilesional; 2 recommended to be performed in parallel; 3 in 30-50% of MMP sera; 4 commercially available (for IgG); 5 only available in specialized laboratories as in-house assays; 6 only with positive direct and/or indirect IF microscopy. only studies with more than 3 patients are indicated; 2 review of Japanese cases; 3 when 17 of the 31 reported sera were re-analyzed by the Biochip ® -based indirect IF assay only 4 reacted with laminin 332. Of these 4 sera, 2 had a malignancy (60). As such, here, only latter data were included; 4 some patients may have also been included in other studies listed here; 5 data of this study were not included in the total numbers since all patients also appeared in the study of the same group by Qian et al. (40). Total numbers are shown in bold. calculated the risk for malignant neoplasms in anti-laminin 332 MMP to be 6.8-fold higher compared to the general population (41). In the recent review by Shi et al., the most frequent tumor in 84 malignancy-associated anti-laminin 332 MMP patients retrieved from the literature, were lung carcinomas (in 23% of patients) followed by gastric (in 17%), uterine (in 13%), pancreatic (8%), colon (8%), ovary (7%), prostate (5%), and thyroid carcinoma (5%) (77). No relation between the recognized laminin chain and the tumor entity was found (77). Of the 12 malignancy-associated antilaminin 332 MMP patients reported by Goletz et al., 3 (25%) had a lung and 2 (17%) a uterine/cervix carcinoma compatible with the data reported by Shi et al., while 2 (17%) revealed a urothel carcinoma and none has a gastric malignancy (38,77). These data suggest that in anti-laminin 332 MMP, solid malignancies predominate with lung and uterine/cervix cancers being among the most prevalent entities, while the distribution of other solid malignancies may also depend on the population. Interestingly, in patients with serum reactivity against a6b4 integrin, no higher rate of malignancies was found alike in MMP patients in general irrespective of the target antigen (78)(79)(80). The exact reason for the association of ani-laminin 332 reactivity and solid cancers has not been fully elucidated yet. It is well known that laminin 332 is relevant for tumor proliferation and migration (81-83). Some solid tumors may produce excessive amounts of laminin 332 and an imbalance of extracellular matrix proteins including laminin 332 was shown to promote tumor cell migration via the Pi3-akt pathway as well as the differentiation of tumor-associated fibroblasts and tumor angiogenesis (84)(85)(86). As such, it may be hypothesized that an imbalance in laminin 332 expression during carcinogenesis induces an autoimmune response that leads to laminin 332-specific autoimmunity including antilaminin 332 antibodies (87)(88)(89). This view is supported by the observation that MMP can regress after excision of the tumor (87,90,91). Pathophysiology of anti-laminin 332 pemphigoid Preliminary evidence for the pathogenic relevance of antilaminin 332 IgG stems from the intraindividual correlation of anti-laminin 332 IgG serum levels with disease activity (38). Apart from in-vitro organ culture models of MMP employing normal human conjunctiva (25, [92][93][94], two mouse models of anti-laminin 332 MMP have been developed. One model reflects the inflammatory-poor variant of MMP and lesions develop independently of complement activation and the infiltration of inflammatory cells in the tissues, while the other model shows, oral, conjunctival, and skin lesions with inflammatory infiltrates and requires the involvement of the Fcg-receptor and activation of C5aR1 (95)(96)(97)). In latter model, dapsone has recently been shown to be effective supporting the notion that this model recapitulates important features of the human disease (98). Because most recent publications used the latter mouse model, a detailed description is depicted in Figure 5. In line with previous findings, methylprednisolone as another first-line therapy for MMP, was also able to reduce the severity of skin, although not oral lesions in this mouse model. In this study, Ghorbanalipoor et al. also showed that parsaclisib, a selective inhibitor of phosphoinositide 3-kinase delta (PI3Kd) significantly reduced skin and oral mucosal lesions (99,100). With regard to the characteristic symptom of scarring, typically occurring at the eyes of anti-laminin 332 MMP patients, this mouse model may also be suitable to unravel signaling pathways that contributes to this specific immunopathogenesis. Biopsies of the palpebral conjunctiva and the skin collected 28 days after the initiation of this model revealed highly condensed collagen fibrils in picro-sirius red staining and trichrome histological staining. In addition, biochemical analysis provided results on altered collagen-crosslinking signaling pathways in these tissues that are associated with fibrosis (101). Furthermore, the previously published upregulation of aldehyde dehydrogenase (ALDH1) in conjunctiva and in fibroblasts isolated from MMP patients with severe eye involvement, could be verified by transcriptome analysis of perilesional skin from this model (102). The inhibition of ALDH1 by disulfiram decreased disease severity in a mouse model for allergic eye disease (102). However, disulfiram was not effective in the anti-laminin 332 mouse model. Here, the same dosage and application of disulfiram was not able to reduce the severity of the conjunctival lesions (101). In-vitro models specific for anti-laminin 332 pemphigoid are rare. Recently Bao et al. published results about anti-laminin 332 MMP patient antibodies that were sufficient to release inflammatory mediators upon binding to keratinocytes without the presence of inflammatory cells and as such without the usage of Fc-receptors. Thus, arising the question whether blistering may be a consequence of just the binding of the anti-laminin 332 IgG and whether the complement system has a nonobligatory role in the initiation of the inflammatory response (103, 104). Anti-laminin 332 reactivity in other pemphigoid diseases Outside MMP, antibodies against laminin 332 have been detected in individual patients with bullous pemphigoid, anti-p200 pemphigoid, and epidermolysis bullosa acquisita in addition to the autoantibodies against BP180, p200 protein, and type VII collagen, respectively (6-11). The report of anti-laminin 332 reactivity in about 40% of bullous pemphigoid sera was not confirmed in subsequent studies (38, 55, 58, 59). When Holtsche et al. investigated the specificities of serum autoantibodies in anti-p200 pemphigoid, antilaminin 332 IgG was observed in 43 (18%) of 239 patients in addition to reactivity against the p200 protein and/or laminin g1 (10). Autoantibody reactivity in the very rare entity orf-induced pemphigoid has puzzled investigators for many years. Recently, Yilmaz et al, showed that the major target antigen in orf-induced pemphigoid is laminin 332 (5). Of note, while a single patient with orf-induced MMP has been described, all other cases associated with orf did not show predominant mucosal involvement and, consequently may be termed orf-induced pemphigoid when antibodies against laminin 332 are detected or orf-induced epidermolysis bullosa acquisita in case of type VII collagenspecific antibodies (5,105,106). The reason why autoimmunity against laminin 332 is not associated with predominant mucosal manifestations when induced by an orf infection is enigmatic. It may be speculated that an underlying molecular mimicry between an orf virus protein and laminin 332 leads to autoantibodies against distinct epitopes on laminin 332 different from those targeted in anti-laminin 332 MMP. Of note, autoantibodies in orf-induced pemphigoid are predominantly of the IgG2 and IgG3 subclasses compared to IgG4 in anti-laminin 332 MMP (38, 107). Conclusion After diagnosis of MMP, testing for serum antibodies against laminin 332 and, when present, a search for the most prevalent solid tumors including chest, abdominal, and pelvic CT, gastroscopy, coloscopy, as well as urological and gynecological examinations appears to be mandatory. The anti-laminin a3 mouse model of MMP may be helpful to decipher key molecules and pathways in the pathophysiology of MMP. Only after definite preclinical data have been generated a randomized controlled treatment study will be initiated and open new therapeutic avenues for patients with this rare and frequently detrimental disorder. Conflict of interest ES has research grants with Admirx, ArgenX, AstraZeneca, Biotest, Dompe, Euroimmun, CSL, Alpine Immune, and Fresenius Medical Care and in the last three years, received consulting fees and/or honoraria from Almirall, ArgenX, AstraZeneca, Janssen, Bristol-Myers Squibb, Chugai, Leo, and Sanofi. SP and ES have a patent application with Dompe. Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. 16	2023-08-12T15:20:26.564Z	2023-08-10T00:00:00.000	{ "year": 2023, "sha1": "9420ac72d8e29cd47732a2180d4dd7d4518af1c5", "oa_license": "CCBY", "oa_url": "https://www.frontiersin.org/articles/10.3389/fimmu.2023.1250115/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "7652f377e69b48d392381250ccd7952450515525", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
212751551	pes2o/s2orc	v3-fos-license	Glucocerebrosidase: Functions in and Beyond the Lysosome Glucocerebrosidase (GCase) is a retaining β-glucosidase with acid pH optimum metabolizing the glycosphingolipid glucosylceramide (GlcCer) to ceramide and glucose. Inherited deficiency of GCase causes the lysosomal storage disorder named Gaucher disease (GD). In GCase-deficient GD patients the accumulation of GlcCer in lysosomes of tissue macrophages is prominent. Based on the above, the key function of GCase as lysosomal hydrolase is well recognized, however it has become apparent that GCase fulfills in the human body at least one other key function beyond lysosomes. Crucially, GCase generates ceramides from GlcCer molecules in the outer part of the skin, a process essential for optimal skin barrier property and survival. This review covers the functions of GCase in and beyond lysosomes and also pays attention to the increasing insight in hitherto unexpected catalytic versatility of the enzyme. Introduction The cellular acid β-glucosidase (EC 3.2.1.45) was first reported to be located in lysosomes more than 50 years ago [1]. There it degrades the glycosphingolipid glucosylceramide (GlcCer), also known as glucocerebroside ( Figure 1A) [2]. The enzyme, commonly named glucocerebrosidase (GCase), is active towards GlcCer molecules with different fatty acyl moieties. Deficiency of GCase causes the recessively inherited disorder Gaucher disease (GD, OMIM #230800, ORPHA355), named after the French dermatologist Ernest Gaucher, who published the first case report [3]. A hallmark of GD are lipid-laden macrophages with lysosomal GlcCer deposits, referred to as Gaucher cells [4]. Numerous mutations in the GBA gene encoding GCase have been associated with GD [5]. The genetic heterogeneity contributes to the highly variable clinical manifestation of the disorder that may involve various organs and tissues [4]. A complete absence of GCase activity is incompatible with terrestrial life due to a disturbed skin barrier [6,7]. The lethal impairment stems from the crucial extracellular role of GCase in the stratum corneum (SC). This review covers the functions of GCase in the metabolism of GlcCer inside lysosomes and beyond. First, Sections 2-5 deal with GCase as a cellular lysosomal enzyme, and in the second part Section 6 onwards focuses on the extracellular function of GCase in the skin. GlcCer is converted by ASAH1 to glucosylsphingosine, Glucosylated cholesterol (GlcChol) formed by GBA2 increases, and GM3 levels rise because increased anabolism by glycosyltransferases to complex GSLs. Enzymes are depicted in green. ASAH1: acid ceramidase, GBA2: cytosolic β-glucosidase, GCase: β-glucocerebrosidase, GCS: glucosylceramide synthase. Glucosylceramide as Intermediate of Glycosphingolipids The primary physiological substrate of GCase is GlcCer, the simplest glycosphingolipid (GSL) in which a single glucose β-glucosidic is linked to the 1-hydroxy of ceramide (Cer) [8]. Figure 2 presents an overview of the GSL metabolism. De novo formation of Cer starts on the endoplasmic reticulum (ER) with formation of 3-keto-dihydrosphingosine by the enzyme serine palmitoyl transferase (SPT) that conjugates the amino acid serine with a palmitoyl chain [9][10][11][12]. Next, the enzyme 3-ketosphinganine reductase (KSR) converts 3-keto-hydrosphingosine to dihydrosphingosine (sphinganine). Ceramide synthases (CERS) are responsible for acylation of dihydrosphingosine, thus generating diverse dihydroceramides [13][14][15]. In mammals six distinct CERS enzymes with different fatty acyl-CoA affinities have been identified. Subsequently, dihydroceramide desaturase (DES) catalyzes the conversion of dihydroceramides into ceramides 15 . Ceramide is alternatively formed in the salvage pathway by acylation of sphingosine molecules released from lysosomes [16,17]. Cer can be further metabolized by conjugation of its 1-hydroxy, resulting in very diverse structures like ceramide 1-phosphate (C1P), sphingomyelin (SM), 1-O-acylceramide, galactosylceramide (GalCer), and GlcCer (reviewed in [18]). Formation of GlcCer, the key GSL of this review, involves transfer of Cer to the cytosolic surface of the Golgi apparatus where the membrane-bound glucosylceramide synthase (GCS) generates GlcCer using UDP-glucose as sugar donor and Cer as acceptor [19,20]. Next, some of the newly formed GlcCer molecules are converted back to Cer by the cytosol facing β-glucosidase GBA2 [21], but most reach via an unknown mechanism the luminal membrane of the Golgi apparatus. There, conversion to more complex GSLs like gangliosides and globosides occurs through stepwise addition of additional sugar and sulfate moieties (the biosynthesis and vast structural heterogeneity of GSL is excellently reviewed in [13,22]). The major destination of newly formed GSLs is the outer leaflet of the plasma membrane. At the cell surface, GSLs fulfill a variety of important functions. GSLs interact with cholesterol molecules via hydrogen bonds and hydrophobic van der Waal's forces and spontaneously form semi-ordered lipid microdomains, commonly referred to as lipid rafts [23,24]. Hydrophilic cis-interactions among GSL headgroups promote lateral associations with surrounding lipid and proteins. Residing in the GSL-enriched domains are proteins involved in interactions of cells with the exterior (extracellular space and other cells) and mediating the associated intracellular signaling processes [24][25][26]. The GSL composition of lipid rafts may exert modulating effects in the cell's response to triggers. One example in this respect is the insulin receptor whose signaling is negatively influenced by neighboring gangliosides such as GM3 in lipid rafts [27][28][29]. Pharmacological reduction of GSLs results in improved glucose homeostasis in obese insulin-resistant rodents [30]. Similarly, the epidermal growth factor (EGF) receptor is influenced by the GSL composition of microdomains in which it resides [31]. GSLs at the cell surface also play direct roles in adhesion/recognition processes. For example, specific GSLs are involved in binding of pathogenic viruses, microorganisms, and bacterial toxins [32,33]. The topic was recently reviewed [34]. Glycosphingolipid-enriched lipid rafts essentially contribute to immunological functions as, for example, activation of T cells [35][36][37][38]. Lysosomal Turnover of Glycosphingolipid GSLs may exit cells from the plasma membrane through incorporation in high-density lipoproteins [39,40]. However, most of the GSLs are internalized from the plasma membrane via endocytosis involving multi-vesicular bodies within late endosomes. Similarly, exogenous GSLs, such as constituents of lipoproteins or components of phagocytosed apoptotic cells, also reach lysosomes by endocytic processes. Upon the delivery of internalized material to lysosomes, fragmentation of GSL components takes place by step-wise removal of terminal sugars by specialized glycosidases. The process is further assisted by accessory proteins such as saposins A-D and GM2 activator protein (reviewed in [41]). The final lipid product of lysosomal fragmentation of GSLs, GalCer, and SM is in all cases Cer [41]. The lysosomal acid ceramidase (EC 3.5.1.23) subsequently splits Cer into free fatty acid and sphingosine to be exported to the cytosol [42]. The cytosolic sphingosine can then be used for the formation of Cer or can be converted by sphingosine kinases (SK1 and SK2) to sphingosine-1-phosphate (S1P) [43]. in the ER, these bind to the membrane protein LIMP2 (lysosomal membrane protein 2) [72][73][74]. This binding is mediated by hydrophobic helical interfaces on both proteins [75]. Action myoclonus renal failure syndrome (AMRF) is a recessively inherited disease caused by mutations in LIMP2 [76]. In most cell types of AMRF patients, except for phagocytic cells, GCase is markedly reduced due to faulty transport to lysosomes [76,77]. More recently, progranulin (PGRN) was identified as another factor influencing GCase [78,79]. PGRN is thought to function as a chaperone facilitating the transport of GCase to lysosomes. It recruits heat shock protein 70 (HSP70) to the GCase/LIMP2 complex in the ER and thus promotes delivery of GCase to lysosomes [80]. Another protein found to interact with newly formed GCase in the ER is ERdj3 [81]. Catalytic Activity of GCase The primary substrate of GCase is GlcCer, as is reflected by the prominent accumulation of this lipid during GCase deficiency [82][83][84]. However, it recently has become apparent that catalytic versatility of the enzymes needs consideration. Firstly, GCase was found able to hydrolyze artificial β-xylosides [20]. Secondly, several retaining β-glycosidases are reported to be able to transglycosylate when provided with a suitable aglycon acceptor ( Figure 1B) [85]. Such catalytic activity has also been observed for GCase, the enzyme being able to generate glucosylated cholesterol (GlcChol) by transglucosylation [86][87][88]. This reaction occurs during cholesterol accumulation in lysosomes as occurs in Niemann-Pick disease type C (NPC) [86]. Massive accumulation of GlcChol in the liver of NPC mice was demonstrated. Inducing lysosomal cholesterol accumulation in cultured cells by their exposure to U1986663A is accompanied by formation of GlcChol [86]. Of note, under normal conditions GlcChol is primarily degraded by GCase into glucose and cholesterol. It may be envisioned that further research will reveal that there exist more β-glucosidic metabolites being substrates (and products) of GCase. Gaucher Disease, a Lysosomal Storage Disorder Since degradation of GSLs is catalyzed by lysosomal glycosidases, inherited deficiencies in these enzymes cause lysosomal accumulation of their GSL substrates, so-called glycosphingolipidoses [9,41,[89][90][91]. Examples of such disorders are Gaucher disease, Krabbe disease, GM2-gangliosidosis, Sandhoff disease, and GM1-gangliosidosis. The glycosphingolipidoses are clinically diverse and generally show marked heterogeneity in severity of disease that usually involves neuropathy in more severely affected patients. Gaucher disease is a prototype glycosphingolipidosis. The first case report was published in 1882 by Ernest Gaucher concerning a female patient with unexplained massive splenomegaly without leukemia [3]. Soon it was recognized that this patient represented a distinct disease entity that was subsequently referred to as Gaucher's disease or Gaucher disease (GD). Following the demonstration of abnormal accumulation of GlcCer in GD patients as the result of deficient GCase activity, the GBA gene encoding the acid β-glucosidase was cloned and characterized [4]. The GBA gene is located at locus 1q21 and neighbored by a pseudogene [92]. Numerous mutations in GBA have now been associated with GD. The consequences of mutations at the level of the GCase protein may markedly differ. For example, the common N370S GCase mutation among Caucasians results in near normal production of a mutant enzyme with aberrant catalytic properties [93]. The heteroallelic presence of this mutation protects against a neuronopathic disease course [4,44,94]. This GBA mutation occurs relatively frequently among Ashkenazim and was proposed to offer some advantage against an infectious disease, for example, bubonic plaque or tuberculosis [95][96][97][98][99]. In contrast, another common pan-ethnic mutation encoding L444P GCase, results in a mutant enzyme that largely misfolds in the ER and consequently only a small fraction (<10% of normal) reaches lysosomes. Homozygosity for the L444P mutation is always associated with a severe neuronopathic disease manifestation [4]. The L444P mutation is thought to have arisen repeatedly by homologous recombination of the GBA gene with its pseudogene. The genetic heterogeneity of GD is accompanied by clinical heterogeneity of the disorder. Common symptoms manifesting in GD patients are hepatosplenomegaly, hematological abnormalities like anemia and thrombocytopenia, skeletal disease, and neuropathology. A very severe manifestation (referred to as collodion baby) involves lethal skin barrier dysfunction [4]. Discrete phenotypic variants of GD are historically discerned: type 1, the non-neuronopathic variant; type 2, the acute neuronopathic variant; type 3, the subacute neuronopathic variant and the collodion baby or neonatal variant. It was proposed to no longer adhere to this classification, but rather view GD as a continuum of phenotypes [99]. Marked intraindividual variation occurs in type 1 GD patients in the nature and degree of organ involvement and particular symptoms such as skeletal disease [4]. The correlation of GBA genotype with GD phenotype is limited in some aspects. While the presence of N370S GCase protects GD patients against neuropathology, there are several reports of monozygotic GD twins with discordant severity of visceral disease [100,101]. A very specific clinical course is associated with the presence of D409H GCase involving yet unexplained cardiac symptoms, including aortic valve, mitral valve, and ascending aorta calcifications [102][103][104]. Modifier genes, and possibly epigenetics and external factors, are considered to impact on the clinical manifestation of GCase deficiency. The transmembrane protein CLN8 (ceroid-lipofuscinosis, neuronal 8), recycling between the ER and Golgi apparatus, is a putative modifier [105]. CLN8, identified as putative modifier of GD in a genome-wide association study, has recently been reported to be involved in the transport of newly formed lysosomal enzymes between ER and Golgi [106]. Other proteins are known to directly influence the life cycle and activity of GCase. Saposin C is the lysosomal activator protein of GCase, and patients with a defective saposin C develop symptoms similar to GD patients [68]. LIMP2, encoded by the SCARB2 (scavenger receptor class B, 2) gene, transporting GCase to lysosomes was reported to be a GD modifier [107]. Polymorphisms in the UGCG gene coding for GCS catalyzing synthesis of GlcCer have also been proposed as GD modifiers [108]. Recently, microRNAs up-or downregulating GCase and downregulating LIMP2 were reported [109]. It has recently been appreciated that carrying a mutant GBA gene is not without health risk. Carriers of GD have a yet unexplained significantly increased (20-fold) risk for developing Parkinson disease (PD) and Lewy body dementia (LBD) [110][111][112]. A recent study in the United Kingdom revealed that 5-25% of patients with PD carry glucocerebrosidase gene mutations, and 10-30% of glucocerebrosidase carriers will develop PD by age 80 [113]. Of note, active GCase activity is also decreased, and corresponding glycosphingolipid substrate levels elevated, in the brain in PD without GBA1 mutations [114,115]. Abnormalities in multiple enzymes and other proteins involved in sphingolipid metabolism were observed in association with PD [114,116,117]. With increasing age, the brain of mice shows reduced GCase levels and increased amounts of lipid substrate [115]. PD is historically viewed as a "proteinopathy" with cellular protein aggregates like that of α-synuclein (αSyn). It has more recently been hypothesized that sphingolipid abnormalities may be primary disturbances that can produce protein aggregation [114]. Indeed, inactivation of GCase promotes accumulation of αSyn aggregates [118]. It was observed that insoluble alpha-synuclein positive aggregates in sporadic PD midbrain linearly correlate with loss of GCase activity [119]. Likewise, protein aggregates develop in mice with primary GBA mutations [120]. Supplementation of GCase or reduction of accumulating glycolipids prevents and reverses α-synucleinopathy [121,122]. It was furthermore observed that over-expression of aggregating αSyn causes a reduction of GCase, suggesting a potential harmful interaction between the two proteins in a self-amplifying manner [123][124][125]. In vitro experiments showed that GCase and aSyn may directly interact at lysosomal pH [126]. Different explanations have been proposed for ways by which mutant GCase may induce α-synucleinopathy (reviewed in [125,127,128]). For example, it was hypothesized that the accumulation of substrates of GCase is pathogenic; that GCase deficiency causes inhibition of autophagy and lysosomal degradative capacity and subsequently reduces turnover of αSyn; that increased αSyn levels impair the activity of GCase and vice versa; and, that GCase deficiency impairs mitochondria. Contrarily, it was proposed that mutant GCase protein may be toxic by inducing an excessive unfolded protein response in the ER or saturating the ubiquitin-proteasome pathway [129,130]. It is conceivable that multiple mechanisms may be involved in the GBA-PD pathology. Lysosomal GlcCer Deposits in Macrophages: Gaucher Cells The storage of GlcCer in GD patients occurs almost exclusively in macrophages residing in the spleen, liver, bone marrow, lymph nodes, and lung ( Figure 1C) [131]. The lipid-laden Gaucher cells are viable, alternatively activated macrophages [132]. These cells overproduce and secrete specific proteins resulting in massively elevated plasma levels in symptomatic GD patients. These proteins are now used as biomarkers of body burden of Gaucher cells. The first identified plasma biomarker is the chitinase named chitotriosidase encoded by the CHIT1 gene [133,134]. It can be conveniently detected by the measurement of its activity towards 4-methylumbelliferyl-chitotrioside [133] and the superior substrate 4-metylumbelliferyl-4 -deoxy-chitobioside [135,136]. Plasma chitotriosidase is on average about 1000-fold elevated in type 1 GD patients. Immunohistochemistry and in situ hybridization revealed that the enzyme is produced by Gaucher cells. Common is a 24-base-pair duplication in the CHIT1 gene that excludes synthesis of active chitinase [137]. The chemokine CCL18/PARC (Chemokine (C-C motif) ligand 18; Pulmonary and activation-regulated chemokine) serves as an alternative plasma marker of Gaucher cells, being 20 to 50-fold elevated plasma of type 1 GD patients [138,139]. The chemokine is over-produced and secreted by Gaucher cells [139]. More recently the glycoprotein nonmetastatic melanoma protein B (gpNMB) was found to be overproduced by Gaucher cells [140]. A soluble fragment of gpNMB is released into plasma and is over 50-fold elevated in type 1 GD patients [140,141]. In cerebral spine fluid and brain of type 3 GD patients elevated gpNMB levels have also been observed [142]. Likewise, recently an increased level of gpNMB in the substantia nigra of PD patients was reported [143]. In mice with conditional deficiency in GCase in the white blood cell lineage Gaucher-like cells are formed. These do not produce chitotriosidase or CCL18, but gpNMB does [140,144]. Inactivation of GCase with an irreversible inhibitor was found to increase gpNMB in the brain [143]. Interestingly, zebrafish and fruit flies overproduce a chitinase during GCase deficiency [130,145]. There is compelling evidence for a direct role of Gaucher cells in GD pathology. Their presence in spleen, liver, and bone marrow is associated with splenomegaly, hepatomegaly, and hematological abnormalities, respectively [4]. The same holds for these symptoms in GD mice with induced GCase deficiency in white blood cells [144]. In GD spleens the storage lesions contain a core of mature Gaucher cells surrounded by pro-inflammatory macrophages [132]. These lesions likely contribute to the complex cytokine, chemokine, and protease abnormalities in GD patients [91,146,147]. Type 1 GD patients show low-grade inflammation and activation of both coagulation and the complement cascade [148,149]. Of note, many of the visceral symptoms of type 1 Gaucher disease patients resemble those of Niemann-Pick type A and B patients suffering from lysosomal acid sphingomyelinase (ASMase) deficiency causing lysosomal sphingomyelin storage [42]. In both disorders, lipid storage in visceral macrophages is a hallmark. In sharp contrast, while GCase is markedly reduced in most cell types of LIMP2-deficient AMRF patients, their symptoms differ from those of type 1 Gaucher patients. Likely this is due to the fact that macrophages of AMRF patients contain a high residual GCase and consequently no lipid-laden macrophages are formed [77]. Therapies of Gaucher Disease: ERT, SRT, PCT/EET The prominence of lipid-laden macrophages in GD and their relationship to pathology has prompted the design of rational therapies aiming to prevent and/or correct the lipid-laden macrophages. The first effective treatment designed for type 1 GD is enzyme replacement therapy (ERT) aiming to supplement patient macrophages with lacking enzyme by repeated intravenous enzyme infusion [150]. Therapeutic GCase, nowadays recombinant but initially isolated from placenta, has enzymatically modified N-linked glycans with terminal mannose groups to favor uptake via the mannose receptor (or another mannose-binding lectin) at the surface of tissue macrophages. Two weekly ERTs reverse hepatosplenomegaly and hematological abnormalities in type 1 GD patients [67]. In addition, it reduces storage cells in the bone marrow [151]. Present ERT does, however, not prevent neurological symptoms due to the inability of enzyme to pass the blood brain barrier. An alternative GD treatment is substrate reduction therapy (SRT) [152][153][154]. SRT aims to balance synthesis of GlcCer with reduced GCase activity of GD patients. Oral inhibitors of GCS (Miglustat and Eliglustat) are approved drugs. Eliglustat therapy resembles ERT in efficacy [155]. Brain-permeable inhibitors of GCS are presently designed and tested [156]. The response to treatment of GD patients is primarily monitored by clinical assessments. A retrospective evaluation revealed that reductions in plasma chitotriosidase during ERT correlate with corrections in liver and spleen volumes, improvements in hemoglobin, platelet count, and bone marrow composition [157]. Given the observed positive outcome of bone marrow transplantation in type 1 GD patients, genetic modification of hematopoietic stem cells was, and still is, seriously considered as therapeutic avenue [144]. At present there is still an unmet need for neuronopathic GD. Small compounds are actively studied as potential therapeutic agents in this respect. One envisioned approach is pharmacological chaperone therapy (PCT). Chemical chaperones are small compounds improving folding of mutant GCase in the ER, thus increasing lysosomal enzyme levels. Current studies with ambroxol, a weak inhibitor of GCase, indicate impressive reductions in spleen and liver volumes in ambroxol-treated type 1 GD patients as well as clinical improvements in type 3 GD patients [158][159][160]. Another approach is enzyme enhancement therapy with small compounds (EET). An example of this is arimoclomol, a heat shock protein amplifier, found to improve refolding, maturation, and lysosomal activity of GCase in GD fibroblasts and neuronal cells [161]. Formation of Glucosylsphingosine From Accumulating GlcCer Important metabolic adaptations occur during GCase deficiency in lysosomes ( Figure 1D) [162]. We demonstrated that part of the accumulating GlcCer is actively converted by lysosomal acid ceramidase to glucosylsphingosine (GlcSph) [163]. GlcSph is sometimes also referred to as lyso-GL1 or lyso-GB1. It was earlier observed that GlcSph is increased in the brain and spleen of GD patients [164,165]. We firstly reported an average 200-fold increased GlcSph level in plasma of symptomatic type 1 GD patients [166]. Urine of GD patients also contains increased GlcSph isoforms [167]. Pharmacological inhibition of GCase in cultured cells and zebrafish embryos causes a rapid increase in GlcSph [168]. The quantitative detection of GlcSph in biological samples was improved by o-phthaldialdehyde (OPA) derivatization and high-performance liquid chromatography [169]. Further improvement was reached by the introduction of LC-MS/MS (liquid chromatography-mass spectrometry) employing an identical (13)C-encoded glucosylsphingosine standard [168]. Measurement of elevated plasma GlcSph is now regularly used in the confirmation of GD diagnosis. Excessive GlcSph in GD patients is believed to contribute to various symptoms. GlcSph was linked to the common reduced bone mineral density (osteopenia) in GD patients by impairing osteoblasts [170]. It is reported to promote α-synuclein aggregation, a hallmark of Parkinson disease [171]. Antigenicity of GlcCer, and possibly GlcSph, is thought to cause the common gammopathies in GD patients, gammopathies that can lead to multiple myeloma [172]. The same lipids were proposed to activate the complement cascade activation and associated local tissue inflammation [173]. GlcSph is hypothesized to diminished cerebral microvascular density in mice, based on the observed interference of the lipid with endothelial cytokinesis [174]. Earlier studies have provided evidence that GlcSph promotes lysis of red blood cells, impairs cell fission during cytokinesis, damages specific neurons, interferes with growth, and activates pro-inflammatory phospholipase A2 (see for a review [91]). In line with these observations is the occurrence of hemolysis, multinucleated macrophages, neuropathology, growth retardation, and chronic low-grade inflammation in GD patients [4]. Of note, in the brain of ageing mice reduction of active GCase in combination with increased glucosylceramide and glucosylsphingosine levels were observed [116]. Excessive Gangliosides In GD patients increases of the ganglioside GM3 (monosialodihexosylganglioside) in plasma and spleen were observed [183]. It is unknown whether this abnormality is caused by increased metabolic shuttling of newly formed GlcCer to gangliosides and/or impaired recycling of gangliosides. Not surprisingly (see Section 2.1), the elevated concentrations of GM3 in GD patients are accompanied by insulin insensitivity, without overt hyperglycemia [184]. Increased Activity of Cytosol-Faced GBA2 and GlcChol Besides GCase, cells contain another retaining β-glucosidase that metabolizes GlcCer. The enzyme GBA2 was discovered during studies with GCase-deficient cells [21]. GBA2 is synthesized as soluble cytosolic protein that rapidly associates to the cytosolic leaflet membranes with its catalytic pocket inserted in the lipid layer. GBA2 shows prominent transglucosylase capacity and is largely responsible for the (reversible) formation of GlcChol from GlcCer and cholesterol [86]. The GBA2 gene (locus 1p13) was identified and GBA2-deficient mice have meanwhile been generated [185,186]. The animals develop normally without overt abnormality, except for incidences of male infertility [185]. GBA2-deficient zebrafish also develop normally [168]. Inhibition of GBA2 in GD and NPC patients treated with N-butyldeoxynojirimycin causes no major complications, whereas on the other hand, individuals with spastic paraplegia and cerebellar ataxia were found to be GBA2 deficient [187][188][189][190]. The physiological role of the highly conserved GBA2 is still an enigma [191]. Reducing GBA2 activity, genetically or using small compound inhibitors such AMP-DNM, has remarkable beneficial effects in NPC mice, ameliorating neuropathology and prolonging lifespan significantly [52,53]. A comparable neuro-protective effect of the iminosugar AMP-DNM was also observed in mice with Sandhoff disease, another neuropathic glycosphingolipidosis [54]. Presently zebrafish models are used to study the poorly understood interplay between GCase and GBA2-mediated metabolism of GlcCer [168]. The possible toxic effect of excessive glucosylated metabolites generated by GBA2 during GCase deficiency warrants further investigation. GCase: Other Locations Than Lysosomes As discussed in Section 3.1, GCase does not rely on mannose-6-phosphate receptor-mediated intracellular sorting and re-uptake after secretion. The intracellular transport of GCase is tightly governed by the membrane protein LIMP2 and secretion of GCase into the extracellular space is normally prevented [77]. Immuno-electron microscopy has revealed that specific organelles are involved in trafficking of GCase-LIMP2 complexes from the Golgi apparatus to lysosomes [192]. The delivery of GCase to other locations than lysosomes warrants consideration and discussion. Lysosome-Related Organelles To fulfill specific physiological functions several cell types have adapted their endolysosomal apparatus and evolved specialized secretory compartments, the lysosome-related organelles (LROs) (for reviews see [193,194]). The LROs are diverse and comprise endothelial cell Weibel-Palade bodies, cytotoxic T cell lytic granules pigment cell melanosomes, and platelet dense and alpha granules. Common components of LROs are tetraspanin CD63, and GTPases RAB27A or RAB27B. The same proteins also occur in multivesicular endosomes (MVEs) that excrete intraluminal vesicles (ILVs) as exosomes upon fusion with the plasma membrane [195]. The notochord vacuole in the zebrafish is also considered to be an LRO [196,197]. Interestingly, LIMP2, the GCase transporter protein, was implicated in the formation of this LRO [198]. An established link between GSLs and LROs concerns the pigmented melanosomes in melanocytes. The formation of melanosomes requires GSLs: melanoma cells when deficient in GCS lose pigmentation due to aberrant transport of the enzyme tyrosinase synthesizing melanin [199]. Similarly, cultured melanocytes lose pigmentation when treated with a GCS inhibitor (Smit and Aerts, unpublished observations). Keratinocytes contain a special kind of LRO, the lamellar body (LB), which justifies more detailed discussion regarding GSLs and their metabolism (see Sections 8 and 9). Prior to this, the composition of the mammalian skin is introduced in the section below. Skin Differentiation and Barrier Formation The mammalian skin acts as a key barrier offering protection against xenobiotics and harmful pathogens and preventing excessive water loss from the body (Figure 2A) [200]. The barrier function resides in the epidermis, the outermost part of the skin that consists of four distinct layers: the stratum basale (SB), stratum spinosum (SS), stratum granulosum (SG), and stratum corneum (SC) [201]. The innermost SB, SS, and SG are the vital parts of the epidermis (thickness: 50-100 µm) while the SC is the non-vital differentiation product (thickness: 10-20 µm). The SB contains proliferating keratinocytes that after escape from this single cell layer start to differentiate and migrate towards the SC, where the keratinocytes differentiate to terminal corneocytes. During this differentiation process the keratinocytes flatten and diminish their water content. During the flattening process cells become filled with keratin. At the interface between the SG and SC, subcellular structures like organelles and nuclei are degraded and corneocytes are formed (as reviewed in [202]). Stratum Corneum: Hydration and Skin-pH Proper function and features of the SC are dependent on optimal water content and acidity. The SC hydration level depends on multiple factors such as amino acids, specific sugars and salts, referred to as the natural moisturizing factor (NMF) [203]. Amino acids of the NMF are breakdown products of the major SC protein filaggrin. Mutations in the filaggrin gene FLG cause a reduced NMF level associated with dry skin [204,205]. NMF also plays a key role in maintenance of pH in the SC. At the outside of the SC the pH is 4.5-5.3 and it gradually increases to pH 6.8 in the inner SC [206]. The local pH likely modulates the activity of various enzymes in the SC, including GCase and ASMase, with optimal catalytic activity at a more acid pH, and thus also impacts on lipid structures [207]. Stratum Corneum: Composition The SC has a "brick-and-mortar" like structure, where the corneocytes are the "bricks" embedded in a lipid matrix that is the "mortar" of the SC [206,208]. During the terminal differentiation of corneocytes, plasma membranes develop into the cornified lipid envelope, a lipid-linked crosslinked protein structure [209]. The cornified lipid envelope acts as template for the formation and organization of extracellular lipid lamellae [210,211]. The lipid matrix contains approximately on a total lipid mass basis 50% ceramides, 25% cholesterol, and 15% free fatty acids with very little phospholipid. The adequate balance of lipid components is essential for proper lipid organization and SC barrier competence [212]. Alterations in the lipid composition have been associated to various skin diseases, particularly to psoriasis, atopic dermatitis and several forms of ichthyosis [213][214][215][216][217][218]. Role of Lamellar Bodies Keratinocytes having specific ovoid-shaped LROs with a diameter of about 200 nm are called lamellar bodies (LBs), or alternatively lamellar granules, membrane-coating granules, cementsomes, or Odland bodies [219]. LBs have a bounding membrane surrounding lipid disks. The main lipids packed in LBs are precursors of ceramides and fatty acids constituting the lamellar matrix in the SC. In the uppermost granular cells, the bounding membrane of the LB fuses into the cell plasma membrane, and the lipid disks are extruded into the intercellular space between the SC and SG. The initially extruded content of the LB is largely metabolized to ceramides and fatty acids and rearranged to form together with cholesterol the intercellular lamellae of the SC. Keratinocytes serve as the initial factory of the permeability barrier of the skin [219]. Briefly, the generation of SC barrier lipids initiates in keratinocytes, where ceramides are de novo formed by ceramide synthase 3 (CerS3). The sphingolipid content of keratinocytes increases along with differentiation. Newly formed ceramides are rapidly modified into glucosylceramides (GlcCers) and sphingomyelins (SMs), thereby likely protecting keratinocytes from cytotoxic ceramide effects. Next, these sphingolipids are packaged into LBs [212]. The membrane protein ABCA12 (ATP-binding cassette sub-family A member 12) is essential for the presence of GlcCer in LBs [220][221][222]. Several mutations in the ABCA12 gene cause Harlequin-type ichthyosis, characterized by thickened skin over nearly the entire body at birth and causing early death. Incorporated in LBs besides lipids are also acid hydrolases including GCase, ASMase, and phospholipase A as well as proteases and antimicrobial peptides. Following exocytotic secretion of LBs, the SM and GlcCer molecules are largely enzymatically re-converted to ceramides [223,224]. Chemical Composition of Skin Sphingolipids The sphingolipids in the skin differ in their complexity of chemical composition from those encountered in most tissues. Firstly, their sphingosine backbones are modified to yield from dihydroceramide (DS) precursors not only the regular ceramide (S) but 6-hydroxyceramide (H), phytoceramide (P) and 4,X-dihydroxysphinganine containing ceramide (T), as well [225][226][227][228]. In addition, skin ceramides have unique fatty acyl moieties. Besides regular non-hydroxylated fatty acyls of variable chain length, there are α-hydroxylated and ω-esterified structures (acylceramides) [229]. In keratinocytes, fatty acids can be elongated by elongases (mainly ELVOL1, ELVOL4, and ELVOL6) [230,231]. Very long chain fatty acids are incorporated in phospholipids and sphingolipids are packaged in LBs. Cholesterol does not require a conversion to be transported into LBs. Cholesterol can furthermore be metabolized to oxysterol or cholesterol sulfate. Oxysterol and cholesterol sulfate can both stimulate keratinocyte differentiation, additionally, cholesterol sulfate has a key role in [232][233][234][235]. Since cholesterol sulfate is highly amphiphilic it can cross the cell membrane and directly enter the SC, where it is metabolized by LB-derived steroid sulfatase to cholesterol [236,237]. Because cholesterol sulfate inhibits proteases that are involved in desquamation [238], its decrease in the upper layers of the SC results in the initiation of desquamation [239,240]. Besides the presence of regular ceramides, the scaffold of the lipid matrix in the SC is built of acylceramides, containing ω-hydroxylated very long chain fatty acids acylated at the ω-position with linoleic acid [212,228]. Also, the acylceramides are synthetized in the keratinocytes, where they and regular ceramides are glucosylated at Golgi membranes and secreted via LB secretion. Extracellularly the linoleic acid residues are replaced by glutamate residues at proteins exposed on the surface of corneocytes, thus completing the corneocyte lipid envelope [212,228,241]. GCase: Crucial Extracellular Role in the Skin Inhibition of either cholesterol, phospholipid, ceramide or glucosylceramide synthesis prevents the delivery of lipids into LBs, disrupting LB formation, thereby impairing barrier homeostasis ( Figure 3) [242]. LB secretion and lipid structure is abnormal in the outer epidermis of multiple skin diseases, like Atopic Dermatitis and Netherton syndrome [215,243,244]. A complete lack of GCase results in a disease phenotype (collodion baby) with fatal skin abnormalities and inhibition of GCase activity reduces the permeability barrier formation [245][246][247][248]. Gaucher mice homozygous for a null allele develop skin abnormalities that are lethal within the first day of life [6,7]. Holleran and colleagues showed increased trans-epidermal water loss (TEWL) and altered barrier function in GCase-deficient mice [248], suggesting deficient conversion of GlcCer to ceramides by GCase alters the skin barrier function. Identical changes were observed in hairless mice treated with GCase inhibitor bromoconduritol B epoxide, however, ceramide levels remained normal [246,248]. Similarly, mice deficient for prosaposin, and therefore also lacking the GBA activator protein saposin C, accumulate GlcCer in the SC and show abnormal SC lamellar membrane structures [249]. Interestingly, deficiency of LIMP2 in AMRF patients is not associated with skin abnormalities. No prominent abnormalities have also been noted in LIMP2-deficient mice. Apparently, GCase is reaching the SC sufficiently without its regular transporting protein. GlcCer and GBA appear to be co-localized in the LB [250][251][252]. GBA activity has been observed throughout the outer parts of the epidermis [253][254][255], and recently a novel in situ method with the use of activity-based probes (ABPs) confirmed predominant localization of active GBA in the extracellular space of the SC lipid matrix [256]. GD is not the only lysosomal storage disease associated with skin barrier abnormalities. In Niemann-Pick disease a deficiency in ASMase causes an impaired conversion of SM into ceramides in the SC and, therefore, into a disturbed skin barrier [248,257]. Reduction of epidermal ASMase activity by the inhibitor imipramine causes delayed permeability barrier repair after SC injury [258]. Atopic Dermatitis A common skin disease is atopic dermatitis (AD, OMIM #603165). Clinical manifestation of AD involves eczematous lesions as well as erythema, xerosis, and pruritis [259][260][261]. In AD there is a complex interplay between inflammation, genetic background, and the skin barrier. Inflammation can affect the skin barrier, and subsequent entry of compounds promotes an immune response. Additionally, it was observed that AD is associated with loss of function mutations in the filaggrin gene FLG [262,263]. As discussed in Section 7.2, filaggrin is essential for SC hydration and may affect the sensitivity of the skin [264]. Even though FLG mutations have been suggested as a predisposing factor for AD, they do not influence SC ceramide synthesis [264][265][266]. SC Lipids in AD SC lipid metabolism and composition have been substantially studied in AD, however there is some disagreement in literature about the lipid composition in the skin of AD patients. Farwanah and co-workers reported no change in non-lesional AD skin compared to control [267], although other studies report a decrease in total ceramide level, as well as an increase in ceramide (AS) and a decrease in ceramide (EOS) and (EOH), mainly in lesional AD skin compared to control [215,266,[268][269][270][271]. Additionally, Di Nardo et al. have reported a decrease in ceramide/cholesterol ratio in AD skin [271]. Besides subclass composition ceramide chain length has also been studied in AD. Some report an increase of short chain ceramides (total chain length of 34 carbon atoms) in lesional AD skin that also correlated with an increased TEWL [268,272]. Moreover, levels of ω-O-acyl-ceramides correlated negatively with TEWL [268]. A reduction in ω-O-acyl-ceramide in AD compared to control was also reported by Jungersted et al. They additionally observed no statistical difference between their FLG mutant and wild-type group in relation to the ω-O-acyl-ceramide decrease [266]. Data on fatty acids in relation to AD skin are limited, but there are a few reports reporting a reduced fatty acid chain length [273,274]. A study by van Smeden et al. described an increase of Atopic Dermatitis A common skin disease is atopic dermatitis (AD, OMIM #603165). Clinical manifestation of AD involves eczematous lesions as well as erythema, xerosis, and pruritis [259][260][261]. In AD there is a complex interplay between inflammation, genetic background, and the skin barrier. Inflammation can affect the skin barrier, and subsequent entry of compounds promotes an immune response. Additionally, it was observed that AD is associated with loss of function mutations in the filaggrin gene FLG [262,263]. As discussed in Section 7.2, filaggrin is essential for SC hydration and may affect the sensitivity of the skin [264]. Even though FLG mutations have been suggested as a predisposing factor for AD, they do not influence SC ceramide synthesis [264][265][266]. SC Lipids in AD SC lipid metabolism and composition have been substantially studied in AD, however there is some disagreement in literature about the lipid composition in the skin of AD patients. Farwanah and co-workers reported no change in non-lesional AD skin compared to control [267], although other studies report a decrease in total ceramide level, as well as an increase in ceramide (AS) and a decrease in ceramide (EOS) and (EOH), mainly in lesional AD skin compared to control [215,266,[268][269][270][271]. Additionally, Di Nardo et al. have reported a decrease in ceramide/cholesterol ratio in AD skin [271]. Besides subclass composition ceramide chain length has also been studied in AD. Some report an increase of short chain ceramides (total chain length of 34 carbon atoms) in lesional AD skin that also correlated with an increased TEWL [268,272]. Moreover, levels of ω-O-acyl-ceramides correlated negatively with TEWL [268]. A reduction in ω-O-acyl-ceramide in AD compared to control was also reported by Jungersted et al. They additionally observed no statistical difference between their FLG mutant and wild-type group in relation to the ω-O-acyl-ceramide decrease [266]. Data on fatty acids in relation to AD skin are limited, but there are a few reports reporting a reduced fatty acid chain length [273,274]. A study by van Smeden et al. described an increase of shorter fatty acids, mainly saturated fatty acids with 16 and 18 carbon atoms, as well as a reduction in fatty acids with 24 carbons or more in non-lesional AD patients [273]. However, another study observed an increased level of very long fatty acid chains in non-lesional as well as lesional AD [274]. It was hypothesized that SC ceramides and fatty acids share a common synthetic pathway, and this is consistent with the observation that ceramide composition is paralleled by the chain length of fatty acids [275]. The expression of enzymes involved in the biosynthesis of fatty acids and ceramides was related to the SC lipid composition in lesional AD skin [276]. Danso et al. observed an altered expression of GBA, ASMase, and CerS3 in lesional AD skin with a corresponding increase in ceramide (AS) and (NS) and decrease in esterified ω-hydroxy CERs. Additionally, they noted increased levels of unsaturated fatty acids and reduced levels of C22-C28 fatty acids in combination with an altered expression of stearoyl CoA desaturase (SCD) and elongase 1 (ELOVL1) [276]. Potential Role for Glucosylsphingosine in AD Pathology Deficiency of ceramides in the SC is thought to contribute to the dry and barrier-disrupted skin of patients with AD. It was proposed that this deficiency involves a tentative novel enzyme named sphingomyelin-glucosylceramide deacylase, forming sphingosylphosphorylcholine (SPC; lysoSM) and GlcSph from SM and GlcCer. Increased deacylase activity is thought to contribute to reduced formation and subsequent deficiency of ceramide in the AD skin [277]. The deacylase enzyme is considered to be distinct from acid ceramidase as based by apparent isoelectric point [278]. Increased deacylase activity was observed for involved SC and epidermis in patients with AD [279]. Unfortunately, the deacylase has so far not been isolated and characterized. At present it cannot be excluded that the intriguing observations are explained by some neutral ceramidase, a bacterial amidase, or even acid ceramidase that in lipid-laden macrophages of GD patients shows GlcCer deacylase activity. A common symptom in AD is pruritis. It was observed that GlcSph induces scratching in mice and more recently it was demonstrated that GlcSph activates the Serotonin Receptor 2 a and b, considered to be part of a novel itch signaling pathway [280,281]. Direct Role of GCase in AD? As discussed above, GCase expression was found to be altered in (particularly lesional) AD skin [276]. However, no abnormality in GCase activity level in AD skin was previously noted [282]. Earlier research in mice pointed to changes in location of GCase activity in mice with a skin barrier disruption [246]. Using the specific and sensitive ABP technology, the localization of active GCase molecules in AD skin has been studied. An abnormal GCase localization in (mainly lesional) AD skin was observed together with abnormal SC lipids (Boer, submitted for publication). It will be of interest to comparably study other skin diseases. It should be stressed that abnormalities in GCase are not a sole cause for AD, however, an acquired local abnormal enzyme activity might contribute to the pathology. Summary and Conclusions This review addresses the multiple functions of the enzyme GCase that degrades the ubiquitous glycosphingolipid GlcCer. In the first part of the review, the metabolism and various functions of glycosphingolipids in health and disease are discussed. The structural features and catalytic mechanism of GCase are described, as well as its remarkable life cycle involving LIMP2-mediated transport to lysosomes. The essential cellular role of GCase in turnover of GlcCer in lysosomes is illustrated by the lysosomal storage disorder Gaucher disease (GD), which results from an inherited GCase deficiency. The review describes the variable symptoms of GD patients and the presumed underlying pathophysiological mechanisms. In addition, it addresses the presently available treatments of visceral manifestations of GD. In the second part of the review, attention is focused on another, extracellular, role of GCase in the skin. In the stratum corneum, GCase converts secreted GlcCer to ceramide, an essential component of lipid lamellae contributing to the barrier properties of the skin. A major lack of GCase activity causes a lethal skin pathology, the collodion baby. To conclude, the catalytic ability of the enzyme GCase has been exploited in evolution for two different functions: in lysosomes, it essentially contributes to cellular glycosphingolipid metabolism, and in the extracellular space of the stratum corneum, it generates an essential building block for lipid lamellae. Conflicts of Interest: The authors declare no conflicts of interest	2020-03-19T10:55:24.087Z	2020-03-01T00:00:00.000	{ "year": 2020, "sha1": "462c12fcd1c0f00b5ef889a4befba39027bdb2ee", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2077-0383/9/3/736/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "73063ab7f7cd85f33d5928a29f31f48890155f75", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
260710806	pes2o/s2orc	v3-fos-license	Sparse Feature Selection for Classification and Prediction of Metastasis in Endometrial Cancer: Extended Abstract Metastasis via pelvic and/or para-aortic lymph nodes is a major risk factor for endometrial cancer. Lymph-node resection ameliorates risk but is associated with significant co-morbidities. Incidence in patients with stage I disease is 4-22% but no mechanism exists to accurately predict it. Therefore, national guidelines for primary staging surgery include pelvic and para-aortic lymph node dissection for all patients whose tumor exceeds 2cm in diameter. We sought to identify a robust molecular signature that can accurately classify risk of lymph node metastasis in endometrial cancer patients. We introduce a new feature selection algorithm, lone star, for applications where the number of samples is far smaller than the number of measured features per sample. We applied lone star to develop a predictive miRNA expression signature on a training. When applied on an independent testing cohort, the classifier correctly predicted 90% of node-positive cases, and 80% of node-negative cases (FDR= 6.25%). Our results indicate that the evaluation of the quantitative sparse-feature classifier proposed here in clinical trials may lead to significant improvement in the prediction of lymphatic metastases in endometrial cancer patients. INTRODUCTION Incidence of pelvic and para-aortic node metastasis in patients with stage I endometrial cancer varies from 4-22% depending on grade, depth of invasion, lymphovascular space invasion, and histologic subtype [6]. Patients harboring tumors less than 2 centimeters in diameter and with less than 50% myometrial invasion are considered to be at low risk for lymphatic metastasis [12]. In a key clinical study, patients whose tumors violate these criteria were recommended for lymphadenectomy. Yet, within this high risk group, only 22% had lymph node metastasis, suggesting that 78% of the lymphadenectomies were unnecessary [12]. A more recent study [11] that separately considered pelvic versus paraaortic lymph node invasion showed little improvement in this statistic. It is therefore clear that current best practice clinical-pathologic parameters are grossly insufficient for reasonable prediction of metastatic disease [12]. In this paper, we develop a new classification algorithm which combines the best aspects of the 1-norm SVM of [5] and the Elastic Net formulation of [16], which uses a convex combination of the 1-norm and the 2-norm squared. We applied this new algorithm to quantitative genome-scale microRNA expression data from 86 clinically annotated pri-mary endometrial tumors, 18 micro-RNAs were recovered that are sufficient to predict the risk of lymph node metastasis within the training cohort. This biomarker panel was tested on an independent cohort of 28 tumors, and returned predictions with high sensitivity, low false discovery rate, and P < 0.0004. The panel therefore provides a path towards the development of a practical molecular diagnostic to avoid unnecessary surgeries (and their associated morbidities) in patients who are not at risk. This study is thus a transdisciplinary combination of two distinct advances: (i) a new algorithm for sparse feature selection in binary classiffication problems, and (ii) its application to predict the risk of metastasis in endometrial cancer. Selection of training cohort and generation of the predictive feature matrix Quantitative measurement of miRNA expression was chosen for detection of putative predictive features. As a family, miRNAs represent a relatively compact feature set which is, never-the-less, profoundly integrated with cell and tissue behavior [3,7,13]. Moreover, miRNA expression patterns have been identified that can predict benign vs. malignant disease, histologic subtypes, survival, and response to chemotherapy [4,10,9]. Two recent surveys highlight the role of miRNAs in cancer in general [1] and endometrial cancer in particular [15]. Total cellular miRNA was extracted from all tissues and measured using LNA-based detection arrays (Supplemental Table 1). 86 samples passed quality controls based on RNA integrity and expression array performance. Among the 1, 428 available probe sets, 213 miRNAs were detectable in all 86 samples. An unsupervised two-way hierarchical clustering of the resulting miRNA expression values within each subclass revealed substantial expression variation between tumors, with no qualitatively evident distinctions between subclasses. Generation of Molecular Signature for Predicting Lymph Node Metastasis In order to detect candidate quantitative microRNA feature sets within the primary tumors that may discriminate between node positive and node negative disease, as well as a numerical procedure for combining the measured values of the features, we turned to machine learning protocols. When the number of features is larger than the number of samples, which is typical for biological problems such as the one here, machine learning approaches commonly encounter a phenomenon known as "over-fitting," wherein a classifier does an excellent job on the training data, but has poor generalization abilities. To overcome this problem, we developed a sparse classification algorithm that uses a convex combination of 1-and 2 norms as a regularization term in its objective function. To detect discriminatory features that may predict metastatic disease, 213 miRNA expression features measured in 86 samples (43 lymph node-positive and 43 lymph node-negative) were used as the training data. The application of the lone star algorithm in the training data with 80 random cross validations at each iteration resulted in a set of 18 features. Afterwards, to compute a unique classifier, a single iteration of lone star is run with these 18 features and the 20 best-performing classifiers giving the best cross-validation error were computed (Supplemental Table 4). To have a more robust classifier the weight vectors and thresholds of these 20 classifiers were averaged to arrive at the weight vector and threshold of the final classifier. This classifier was applied to the 86 tumor training cohort, and it classified all 86 tumors correctly. BIOLOGICAL SIGNIFICANCE OF SELECTED BIOMARKERS We carried out an analysis of the various genes that are regulated by the 18 miRNAs in the final feature set. We retrieved data from the miRTarbase database, which comprises experimentally validated micro-RNA to target gene interactions in humans. A total of 740 genes were recovered, the vast majority of which are associated with the micro-RNA hsa-mir-155. A recent study suggests that hsamir-155 is over-expressed in endometrial cancer patients visa-vis normal patients [14]. CLASSIFIER VALIDATION WITH AN IN-DEPENDENT COHORT To rigorously test the classifier developed using the lone star algorithm, an independent cohort of primary tumors with known metastatic state was collected. This comprised 28 endometrial cancer samples obtained between 2010 and 2012 under an IRB approved Comprehensive Gynecologic Oncology Tumor Repository protocol. The quality of the classification results were determined with a 2×2 contingency table, and computing the likelihood of arriving at the classifications purely through chance. Pvalues were computed using the Fisher exact test [8] and the Barnard exact test [2] . The P -value of the clasiffication was 0.0004 with the Barnard exact test, and 0.0012 with the less powerful Fisher exact test. CONCLUSIONS In this work, we have developed a novel sparse classification algorithm and applied it to predict risk of lymph node metastasis in endometrial cancer patients. The algorithm produced a weighted classifier, using 18 micro-RNAs, and achieved 100% accuracy on the training cohort. When applied to an independent testing cohort, the classifier correctly predicted 90% of node-positive cases, and 80% of node-negative cases (FDR= 6.25%). The classifier developed in this study was based on molecular measurements from excised tumors. If one could predict the risk of lymph node metastasis on the basis of a biopsy, then the decision to carry out lymphadenectomy or not could be made at the time of excision of the primary tumor. Therefore a useful next step would be to repeat the present study on a cohort of biopsies. Pending the completion of such a study, it is worth noting that a prediction of the risk of metastasis is valuable even if lymphadenectomy is not performed, as it can inform choices for post-resection patient care.	2016-11-29T20:03:44.711Z	2016-10-02T00:00:00.000	{ "year": 2016, "sha1": "7463f7579b6a0854d25401b7b656d5989ccee94d", "oa_license": "CCBY", "oa_url": "https://digitalcommons.wustl.edu/cgi/viewcontent.cgi?article=7066&context=open_access_pubs", "oa_status": "GREEN", "pdf_src": "ACM", "pdf_hash": "7463f7579b6a0854d25401b7b656d5989ccee94d", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Computer Science", "Medicine" ] }
13920781	pes2o/s2orc	v3-fos-license	Altered Heparan Sulfate Structure in Mice with Deleted NDST3 Gene Function* We report the generation and analysis of mutant mice bearing a targeted disruption of the heparan sulfate (HS)-modifying enzyme GlcNAc N-deacetylase/N-sulfotransferase 3 (NDST3). NDST3-/- mice develop normally, are fertile, and show only subtle hematological and behavioral abnormalities in agreement with only moderate HS undersulfation. Compound mutant mice made deficient in NDST2;NDST3 activities also develop normally, showing that both isoforms are not essential for development. In contrast, NDST1-/-;NDST3-/- compound mutant embryos display developmental defects caused by severe HS undersulfation, demonstrating NDST3 contribution to HS synthesis in the absence of NDST1. Moreover, analysis of HS composition in dissected NDST3 mutant adult brain revealed regional changes in HS sulfation, indicating restricted NDST3 activity on nascent HS in defined wild-type tissues. Taken together, we show that NDST3 function is not essential for development or adult homeostasis despite contributing to HS synthesis in a region-specific manner and that the loss of NDST3 function is compensated for by the other NDST isoforms to a varying degree. Heparan sulfate (HS) 2 is produced by most mammalian cells as part of membrane and extracellular matrix proteoglycans (1). The chain grows by the copolymerization of GlcA␤1,4 and GlcNAc␣1,4 residues and undergoes modification by one or more of the four NDST isozymes, which remove acetyl groups from subsets of GlcNAc residues and add sulfate to the free amino groups. In vertebrates, ndst1 and ndst2 mRNA are expressed in all embryonic and adult tissues examined, whereas ndst3 and ndst4 transcripts are predominantly expressed during embryonic development and in the adult brain (2). Most subsequent modifications of the HS chain by O-sulfotransferases and a GlcA C5-epimerase depend on the presence of GlcNS residues, making the NDSTs largely responsible for the generation of sulfated ligand binding sites in HS (3)(4)(5). In vitro, NDST3 differs biochemically from the other NDST isoforms by possessing a high deacetylase activity but very low sulfotransferase activity (2). Many growth factors and morphogens bind to HS. In some cases, HS-proteoglycans are thought to act as co-receptors for these ligands. Studies in Drosophila melanogaster demonstrated that HS is crucial for embryonic development (6) and that the fly NDST ortholog, Sulfateless, affects signaling mediated by Wingless (Wg), Hedgehog (HH), and fibroblast growth factor (FGF) (7)(8)(9). The ability of HS to regulate the activity of morphogens and growth factors is currently best understood for the FGFs. HS was found to be a necessary component of FGF-FGF receptor binding and assembly (10), and global changes in HS expression regulate FGF and FGF receptor assembly during mouse development (11). Due to the multiple developmental processes regulated by the 23 FGFs, including those of the lung, limbs, heart, skeleton, and brain (reviewed in Ref. 12), perturbed HS synthesis results in the generation of FGF-related phenotypes (13,14). The crucial role of HS in morphogen transport and on receiving cells has also been demonstrated for vertebrate HH (15)(16)(17)(18) and PDGF function during embryonic vascularization (19). Mouse mutants made deficient in NDST1 have been characterized, demonstrating a crucial role for this isoform for properly modifying HS during development. In the adult mouse, NDST1 and NDST2 also play important roles in the generation of connective-tissue type mast cells, endothelial cell function, and lipid metabolism (15, 18 -26). In this report, we asked to what extent HS function during development and in the adult vertebrate depends on NDST3 function and to what extent NDST3 contributes to the generation of sulfated HS. Moreover, we wished to examine whether the formation of free amino groups present on heparan sulfate is related to NDST3 activity. We describe that NDST3-deficient mice are born at slightly sub-mendelian ratio, are fertile, and show subtle changes in some hematological parameters and in their behavior. No significant overall changes in HS sulfation could be detected in those mice, but microdissection of the adult brain revealed a region-specific activity of NDST3, leading to changes in HS sulfation in the mutant brain. Mice made deficient in both NDST3 and NDST1 function revealed a role of NDST3 in the proper sulfation of nascent HS, resulting in the complete lack of one disulfated disaccharide product. We thus conclude that, although NDST3 is expressed in various tissues and contributes to HS synthesis, its activity can be substituted by the other NDSTs. EXPERIMENTAL PROCEDURES Targeted Recombination of the ndst3 Gene-The thymidinekinase/neomycin-containing targeting vector was constructed by insertion of loxP sites in intron sequences surrounding exon 2 (the first coding exon) of ndst3, including 327 of 873 amino acids of the open reading frame. The final targeting vector was linearized using SalI before transfection of ES cells. R1 ES cells were grown, transfected, and subjected to neomycin G418 selection. Homologous recombinants were identified by Southern blotting and PCR and transfected with a cre-expressing vector, followed by gancyclovir selection. Four type II recombinants were chosen and injected in C57Bl/6J blastocysts. The mouse line obtained was backcrossed into a C57Bl/6 background for Ͼ10 generations. The primers employed for genotyping were: 5Ј-P1: 5Ј-ggtacccggggatcaattcg-3Ј; P2: 5Ј-ccagaaggctaacactgtaaag-3Ј; P3: 5Ј-gaaagtgaagtctctgggcgg-3Ј; and P4: 5Ј-gcttggatgatttggtcacact-3Ј. Assessment of the significance of the deviation from mendelian inheritance was performed using the Chi-square test. Compound mutant mice were derived from matings with NDST1 (18)-and NDST2 (22)-deficient mice. Histology and in Situ Detection of RNA-Embryos were fixed in 4% paraformaldehyde overnight, dehydrated, embedded in paraffin, and sectioned. Sections were stained with hematoxylin and eosin for histological analysis. For in situ hybridization, 700 base probes against the most variable N-terminal region of ndst1 and ndst3 and a 500-base probe against the ndst2 3Ј-untranslated region were employed (DIG RNA Labeling Kit, Roche Applied Science). Quantitation of apoptosis was performed on paraffin sections of two mutant and two wild-type E12.5 embryos, using the TUNEL Assay Kit (Roche Applied Science). Patched expression was detected using anti-PTC1 antiserum (Acris Antibodies, Hiddenhausen, Germany) and secondary fluorescein isothiocyanate-labeled goat anti-rabbit antibodies (Dianova, Hamburg, Germany) on three mutant and wild-type embryos. Adult Mouse Brain Immunohistochemistry-Bielschowsky stain, Gallays stain, anti-MAC-3, anti-PCNA, and anti-GFAP were employed to detect possible cellular abnormalities in NDST3 mutant brain. Images were taken on a Zeiss Axiophot microscope employing a 10ϫ/0.3, a 20ϫ/0.5, and a 63ϫ/1.25 Zeiss objective and a Leica DFC280 camera. Leica software was used for image capturing and Photoshop 7 software run on Macintosh computers for the generation of figures. Contrast FIGURE 1. Disruption of the ndst3 gene by targeted recombination and ndst3 expression analysis. A, maps of the wild-type ndst3 locus, the Type II "floxed" allele, and a Type I deletion allele, obtained after breeding of Type II mice with ZP3-CRE mice. Lox sites located in intron sequences are shown as triangles, and arrows denote fragment sizes following HindIII restriction. B, PCR analysis. Employing primers p3 and p4, deletion of exon 2 in NDST3 Ϫ/Ϫ mice yielded a 670-bp product, whereas primers p1 and p2 produced a 500-bp amplification product in wild-type mice. Heterozygous mice yield both amplification products. C, Southern blot analysis of DNA. DNA digestion using restriction endonuclease HindIII yielded a 6.0-kbp (wt) and a 4.2-kbp (NDST3 Ϫ/Ϫ ) band. D and E, PCR analysis of the same samples as in C. Employing primers p3 and p4 yielded a 670-bp product from the type I mutant (Ϫ/Ϫ) allele (D), p1 and p2 yielded a 500-bp product from the wild-type (ϩ/ϩ) allele (E). F, semiquantitative reverse transcription-PCR analysis of ndst1-ndst4 expression using a human cDNA panel (Clontech). ndst1 and ndst2 are abundantly expressed, whereas ndst3 expression is strongest in brain, kidney, liver, pancreas, spleen, testis, and thymus. and brightness were adjusted for whole images during figure assembly. Preparation of HS-Mutant and wild-type tissues were pooled, homogenized using an Ultra-Turrax homogenizer (IKA, Germany), digested with 2 mg/ml Pronase in 320 mM NaCl, 100 mM sodium acetate (pH 5.5) overnight at 40°C, diluted 1:3 in water, and applied to a 2.5-ml column of DEAE-Sephacel. After washing the column with 0.3 M NaCl, the glycosaminoglycans were eluted with 1 M NaCl. For disaccharide analysis, the GAG pool was ␤-eliminated overnight at 4°C (0.5 M NaOH, 1 M NaBH 4 ), neutralized with acetic acid until the pH was ϳ6 and applied to a PD-10 (Sephadex G-25) column (Amersham Biosciences). Glycosaminoglycans eluting in the void volume were lyophilized, purified on DEAE as described above, again applied to a PD-10 column, and lyophilized. 100 mg to 1 g of tissue, depending on the source, typically yielded 40 -140 g of GAGs. 10 g of GAG samples were digested using heparin lyases I, II, and III (1.5 milliunits of each in 100-l reactions, IBEX, Montreal, Canada) at 37°C for 1 h, and the resulting disaccharides were separated from undigested chondroitin sulfate using a 3-kDa spin column (Centricon, Bedford, MA). Compositional disaccharide analysis of wild-type and NDST3 E15.5 embryos was then carried out by high-performance liquid chromatography analysis using Carbopac PA1 columns (Dionex). Compositional disaccharide analysis of compound mutant embryos and defined adult brain areas was carried out by liquid chromatography/mass spectrometry (LC/ MS). First, disaccharides were separated on a C18 reverse phase column (0.46 ϫ 25 cm, Vydac) with the ion pairing reagent dibutylamine (Sigma), and eluted species were evaluated using a quantitative mass spectrometric method. Analysis of the disaccharide composition by post-column derivatization with 2-cyanoacetamide (27) or by the LC/MS method gave comparable results. 3 A comparison of the two methods using 0.5 g of commercial porcine heparin showed an error of 2% for abundant disaccharides to 20% for minor species. Cell Proliferation-Cell proliferation was measured using fibroblasts derived from E14.5 embryos. Cells were labeled using 100 mM bromodeoxyuridine in medium for 5h, fixed with 4% paraformaldehyde in phosphate-buffered saline, and detected using anti-bromodeoxyuridine antibodies (Zymed Laboratories Inc.). Analysis of FGF2-dependent MAPK pathway activation was performed using anti-ERK1/2 and anti-3 R. Lawrence, R. Cummings, and J. E. Esko, submitted for publication. phospho-ERK1/2 polyclonal antibodies (Promega, Madison, WI). Fibroblasts derived from the heads of E14.5 wild-type and mutant embryos (n ϭ 4) were cultured in DMEM plus 10% FBS, starved for 20 h in DMEM without FBS, incubated in complete medium, DMEM without FBS, or 10 ng/ml FGF2 in DMEM without FBS for 5 min, and lysed. Analyses were done in duplicates. Behavior-For behavioral tests, 13 male and 13 female NDST3 mutant mice were compared with 12 male and 10 female wt controls. Mice were kept under a 12-h/12-h light dark cycle for 3 weeks before testing began. Food and water was available ad libitum. All procedures and protocols met the guidelines for animal care and experiments in accordance with national and European (86/609/EEC) legislation. General health and neurological status were assessed using a protocol, including tests as described elsewhere (28). Animals were inspected for physical appearance and underwent neurological testing, including acoustic startle, visual placing, grip strength, and reflex functions to ensure that behavioral findings were not the result of deteriorating physical conditions of the animals. The Barrier test was employed to assess spontaneous exploratory behavior, the Open-field test to assess exploration and fear of open spaces, and the Elevated plus-maze, consisting of elevated open stages that mice are reluctant to enter, to assess anxietyrelated behavior. All tests were conducted as blind studies. For a detailed description of behavioral tests see Ref. 29. Data analysis was conducted using the statistical software "R" (The R Project for Statistical Computing, available on the web) using non-parametric statistics. Comparison of two samples was done using the two-sample Wilcoxon test (Mann-Whitney U test). RESULTS Targeted Disruption of ndst3-To study the function of the HS biosynthetic GlcNAc N-deacetylase/N-sulfotransferase (NDST) isozymes in mammalian biology, conditional knockout mice for Ndst3 were generated using the cre-loxP system and homologous recombination in embryonic stem cells (Fig. 1). In the targeting vector, CRE-recombination sequences (loxP sites) were positioned in intron sequences surrounding the second exon of ndst3, which included most of the 5Ј-untranslated region and 327 of 873 amino acids of the open reading frame, containing the signal peptide, cytoplasmic tail, transmembrane region, and part of the catalytic domain (Fig. 1A). Chimeric mice were generated by blastocyst injections of four embryonic stem cell clones. The resulting type II ndst3 mouse line was crossed with ZP3-cre mice, deleting the "floxed" allele in the oocyte and generating mice with a systemic deletion of ndst3 (Type I, Fig. 1B). Type I Ndst3 mice showed only an insignificant deviation from the expected Mendelian distribution (28% NDST3 ϩ/ϩ , 50% NDST3 ϩ/Ϫ , and 22% NDST3 Ϫ/Ϫ , n ϭ 283), were fertile, and appeared normal. Expression of ndst3 in the Mouse and in Human Tissues-To investigate ndst3 expression in adult human tissues, semiquantitative reverse transcription-PCR analysis using cDNA derived from various tissues was conducted (Fig. 1F). Only after 38 cycles of amplification, could ndst3 expression be detected in the brain, kidney, liver, pancreas, spleen, testis, and thymus. This expression pattern was more restricted than that of ndst1 and ndst2. Due to the lack of an isoform-specific anti-NDST3 antibody, ndst3 in situ hybridization was next conducted to detect areas of ndst3 transcription. Strongest ndst3 transcription was detected in cerebellar granule cells, the hippocampus, the brain stem, and the cortex/olfactory bulb (Fig. 2, C and G). ndst1 and ndst2 in situ analysis revealed non-overlapping expression of ndst1 restricted to cerebellar Purkinje cells ( Fig. 2A and inset), whereas ndst2 showed overlapping expression in the granule cell layer (Fig. 2B). In the hippocampus, ndst1-3 were all strongly expressed (Fig. 2, E-G). ndst3 expression in the developing embryo was also analyzed at various stages. In the E10.5 embryonic head, ndst3 expression was detected in trigeminal (V) neural crest tissue (supplemental Fig. S1, A-C). In the E12.5 embryonic skull, ndst3 was still expressed in the trigeminal ganglion and additionally in restricted areas of the fourth ventricle, the metencephalic/myelencephalic part of the rhombencephalon, the developing telencephalon, and the spinal cord (supplemental Fig. S1, D-I). ndst3 expression was found to be more widespread in the E15.5 embryo (supplemental Fig. S1, J-L). Strongest expression was found in neural tissue such as the telencephalon (J), the spinal cord (K), as well as in hind brain (L). ndst3 was also strongly expressed in the developing lung and the frontonasal process that forms much of the face (supplemental Fig. S1, J and K). HS Composition in Total Adult Mouse Brain and Embryo-Heparan sulfate can be depolymerized to constituent disaccharides using a combination of three heparin lyases. The individual disaccharides containing one, two, or three sulfate groups can then be separated and quantitated using high-performance liquid chromatography analysis or by mass spectrometry. Disaccharide analysis of HS derived from E15.5 embryos by high-performance liquid chromatography revealed a slight increase in the amount of non-sulfated UA-GlcNAc and monosulfated UA-GlcNAc6S, whereas the amount of UA-GlcNS and UA2S-GlcNS6S was decreased, demonstrating FIGURE 4. Disaccharide analysis of mutant embryo and various regions of the mouse brain. HS was isolated from microdissected NDST3 mutant and wild-type brain or embryo after samples were digested with heparin lyases. The resulting disaccharides were analyzed by quantitative LC/MS. Values denote the mean % of total disaccharide. A, overall sulfation and relative amounts of disaccharides are highly variable among various wild-type brain regions and the wild-type embryo. Wild-type cortex and hippocampus show high levels of sulfation and wild-type cerebellum, and embryo showed lower levels. B-F, compositional disaccharide analysis in brain stem (pons and medulla, B), cerebellum (C), hippocampus (D), and cortex (E) derived from wild-type and NDST3 mutant mice. Relatively unchanged disaccharide composition in the brain stem and hippocampus (B and D) indicate low NDST3 activity in those tissues, or compensatory activity of other NDST isoforms. However, the relative amounts of mono-, di-, and trisulfated disaccharides in the cortex (E) were reduced, and the relative amount of non-sulfated UA-GlcNAc was strongly increased, indicating that NDST3 contributes to HS synthesis in that brain area. In contrast, an increase in the relative amount of trisulfated UA2S-GlcNS6S and decrease in non-sulfated UA-GlcNAc in the cerebellum (C) upon NDST3 deletion indicates overcompensation by another Ndst isoform, possibly by NDST2, which mediates synthesis of highly sulfated heparin in mast cells and is highly expressed in cerebellar granule cells. Results are presented as percent of total disaccharide. F, quantitative LC/MS results of B-E. Disaccharides UA2S-GlcNAc, UA-GlcNH6S, UA-GlcNH, and UA2S-GlcNH were not detected in any tissue investigated. TABLE 1 Total amount of sulfates per 100 disaccharides in various brain regions Disaccharides were analyzed by quantitative LC/MS, and sulfation was calculated from those results. The highest overall sulfation was detected in hippocampus, cortex, and brain stem; lower sulfation levels were detected in the cerebellum. Generally, the difference in overall sulfation was due to parallel differences in N-, 6-O-, and 2-O-sulfation. In the NDST3 Ϫ/Ϫ brain stem, hippocampus, and cortex, the total amount of sulfation decreased or was unaltered. In the cerebellum, a significant increase in N-, 6-O-, and 2-O-sulfation was observed, possibly by compensatory NDST2 activity that results in increased "heparin-like" HS sulfation. Analysis of NDST3-deficient mice some NDST3 activity in the embryo (Fig. 3). We next analyzed disaccharide composition of purified HS from mutant and wild-type P50 mouse brain by quantitative LC/MS. As shown in Fig. 4A and in Table 1, the amount of sulfation varied in different regions of wild-type mouse brain. Most notably, the cerebellum had reduced levels of all sulfated disaccharides (69 sulfates per 100 disaccharides). The highest overall sulfation was detected in hippocampus, cortex, and brain stem (112, 103, and 96 sulfates per 100 disaccharides, respectively). Generally, the difference in overall sulfation of wild-type tissues was due to parallel differences in N-, 6-O-, and 2-O-sulfation (Table 1). Like in the embryo, NDST3 deletion did not lead to large changes in HS sulfation in various parts of the brain (Fig. 4, B-E), with two notable exceptions. In the cerebellum, the amount of trisulfated disaccharide UA2S-GlcNS6S increased ϳ2-fold, whereas the amount of nonsulfated UA-GlcNAc decreased (Fig. 4C). In the cortex, all of the N-sulfated disaccharides decreased with the exception of UA-GlcNS6S, and UA-GlcNAc increased strongly (Fig. 4E). Detection of disaccharides containing free amino groups was also included in the analysis to determine the role of NDST3 in their generation. UA-GlcNH6S and UA2S-GlcNH were not detected in wildtype or mutant samples (data not shown). UA2S-GlcNH6S was detected in wild-type and mutant cerebellum and hippocampus, and no reduction was noted in NDST3 Ϫ/Ϫ animals (Fig. 4F). These results indicate that deletion of NDST3 in the adult brain results in a variable and region-specific change in sulfation patterns. Brain stem Histology and Immunohistochemistry of Mutant Tissues-Anti-HS HepSS1 and 10E4 antibody stainings were comparable on adult tissue sections and cultured embryonic fibroblasts. HepSS1 staining was detected in all mutant and wild-type tissues at all stages (supplemental Fig. S2, A and B). Despite high ndst3 expression in the embryo, but consistent with only moderate changes in HS sulfation, no significant reduction in FGF2dependent MAPK signaling as judged by ERK1/2 phosphorylation was observed in E14.5 cultured mouse embryonic fibroblasts (wt: 100% Ϯ 9% versus NDST3 Ϫ/Ϫ : 90% Ϯ 19%, p Յ 0.4, n ϭ 4). No reduction in the expression of the hedgehog receptor Patched (PTC) could be observed in the developing mouse head, indicating normal Hedgehog signaling in the mutant (supplemental Fig. S2, A and B). Based on the high ndst3 expression and the altered HS profile in the adult mouse brain, histological analysis of the brain and immunohistochemical analysis of prominent cell types was conducted. Adult mouse brain analysis using Bielschowsky stain (to stain reticular fibers, neurofibrils, axons, and dendrites), Gallays stain (diffuse and neuritic plaques, amyloid in the central core of neuritic plaques and neurofibrillary tangles), anti-MAC-3 (macrophages), anti-PCNA (cell proliferation) and anti-GFAP (glia) did not detect significant differences between the NDST3 mutant and wild-type brains (data not shown). Neurofilament staining at E12.5 was also conducted to investigate whether the observed ndst3 expression in trigeminal neural crest tissue was indicative of NDST3 function in the development of the peripheral nervous system. Again, no difference in the staining of neurofilament-expressing nerves and in the fasciculation of peripheral nerves could be observed between mutant mice and wild-type littermate controls (supplemental Fig. S2, C and D). We thus conclude that, in the NDST3 mutant embryo and in adult brain, both of which normally express high levels of NDST3, no cellular changes occur in NDST3 mutant mice despite a variable change in overall sulfation and disaccharide composition. Immunology, Urine Analysis, and Hematology-Because ndst3 expression was found in adult human and mouse kidney, we next assessed kidney function by urine analysis. No significant changes in the levels of glucose, bilirubin, ketones, blood, protein, nitrites, and pH were found in the mutant mouse. Subsequent urine analysis employing an ApiZym assay also revealed no significant changes. Kidney morphology as assessed by histological analysis did not reveal any change in size or any dysmorphology (not shown). ndst3 expression in the thymus and spleen also prompted us to investigate immunological parameters. Again, cellularity counts of the lymph node, bone marrow, and thymus revealed no significant differences (Fig. 5). The bone marrow cellularity count revealed no significant difference between B cells, T cells, myeloid cells (Mac1), and erythroid cells (Ter119) in wild-type and NDST3 mutant mice. In the thymus, the relative number of CD4 single-positive cells, CD8 single-positive cells, CD4/CD8 double-positive cells, and CD4/CD8 double-negative cells was unchanged. The lymph node cellularity count did not reveal differences between ndst3 mutant mice and wild-type littermate controls (CD4, CD8, B220, B220/B7.2, B220/L-Sel, B220/CD44, Gr1, IgM/IgD, IgD/B220, IgM/B220, CD22/CD21, CD22/B220, CD21/B220, CD23/CD40, CD23/B220, CD40/B220, CD79b/B220, and NK1.1 cells were measured). Further hematological assays showed a slight reduction in Protein C levels (20% decrease, p Ͻ 0.06, n ϭ 10 wt versus 8 mutant mice) (Fig. 6A). However, no significant change in bleeding time (25 s Ϯ 44 s (wt) versus 31 s Ϯ 17 s (mutant)) could be associated with it. We also found reduced levels of total cholesterol (23% decrease, p Յ 0.005) and of high density lipoprotein (31% decrease, p Յ 0.003, n ϭ 10 wt versus 9 mutant mice) (Fig. 6, B and C). Cholesterol reduction was again found in an independently reproduced experiment (20% reduction in females, p Յ 0.001, n ϭ 8 wt mice versus 4 mutants). Taken together, although subtle but significant changes in the levels of Protein C and cholesterol were detected in NDST3 mutant mice, those mutants show otherwise normal hematological parameters. HS Composition in Compound Mutant Embryo-To examine whether NDST1 and NDST3 modify an overlapping set of HS motifs, HS-disaccharide analysis in the compound mutant E16.5 embryos was performed by LC/MS, and the result was compared with each one of the single mutants (Fig. 8). Both NDST1 and NDST3 contribute to UA2S-GlcNS and UA-GlcNS6S production. However, the relative amounts of UA-GlcNS, UA-GlcNAc6S, and UA2S-GlcNS6S were not further reduced in the compound mutant embryo if compared with the NDST1 Ϫ/Ϫ ;NDST3 ϩ/Ϫ embryo. In total, 43.73% sulfated disaccharides were detected in NDST1 ϩ/Ϫ ;NDST3 Ϫ/Ϫ , 31.83% in NDST1 Ϫ/Ϫ ;NDST3 ϩ/Ϫ , and 24.57% in NDST1 Ϫ/Ϫ ; NDST3 Ϫ/Ϫ mutants, indicating partial compensatory activities of NDST1 and NDST3. These results show that, in the embryo, NDST3 deficiency impaired downstream sulfation reactions to a small and varying extent. They also demonstrate preferential activity of NDST3 on UA2S-GlcNS and UA-GlcNS6S containing HS motifs and partially overlapping activities of NDST1 and NDST3. The latter finding was confirmed by the severe phenotypes of NDST1 Ϫ/Ϫ ;NDST3 Ϫ/Ϫ compound mutant embryos, indicating that NDST3 contributes to the development of the skull, brain, and eyes. We also investigated proliferation of isolated E14.5 fibroblasts derived from NDST1;NDST3 compound mutant embryos, NDST3 mutant embryos, and wild-type embryos under normal serum conditions immunohistochemically after bromodeoxyuridine incorporation. Again, no differences could be observed between mutant and wild-type fibroblasts (25% Ϯ 4% proliferating wild-type cells, 27% Ϯ 9% proliferating mutant FIGURE 6. NDST3 mutant mice show a moderate hematological phenotype. A, quantification of protein C activity (% inhibition) present in blood samples of NDST3 Ϫ/Ϫ and wild-type mice. The average value of 107% Ϯ 19% inhibition in wild-type mice was reduced to 86% Ϯ 24% inhibition in mutants (n ϭ 17). B, reduced blood cholesterol levels in NDST3 Ϫ/Ϫ mice. The average amount of cholesterol in wild-type mice (109 Ϯ 21 mg/dl) was significantly higher than the amount in NDST3 mutant mice (84 Ϯ 12 mg/dl, n ϭ 19). C, high density lipoprotein (HDL) levels were reduced in NDST3 Ϫ/Ϫ mice. The average amount of high density lipoprotein in wild-type mice (93 Ϯ 18 mg/dl) was significantly higher than the amount in NDST3 mutant mice (64 Ϯ 18.5 mg/dl, n ϭ 19). Triglyceride levels are comparable in both groups. D, reduced anxiety-related behavior in mutants if compared with the wild-type mice. Reduced anxiety-related behavior in mutants was compared with wild-type mice and measured as the quotient of open arm entries (qno) in the Elevated plus-maze. A significant difference is found in male mice (n ϭ 22) but only a trend in females (n ϭ 26). The mutant forebrain forms a single holosphere. 4, fourth ventricle; t, telencephalon. D, TUNEL staining reveals strongly enhanced levels of programmed cell death in maxillary prominences of E12.5 NDST1;NDST3 compound mutant embryos (right) if compared with NDST3 mutant littermate controls. E, quantification of apoptotic cells found in the cortex, hindbrain (hb), thalamus, and maxillary prominences (mp). Significantly elevated levels of apoptosis are found in hindbrain (ϩ1000%, p Ͻ 0.01) and maxillary prominences (ϩ380%, p ϭ 0.015). Values are expressed as % relative to wild-type levels. DISCUSSION In this report, we show that development and physiology of NDST3 mutant mice were not significantly impaired despite strong ndst3 expression in the embryo and in the adult brain, kidney, liver, pancreas, spleen, testis, and thymus. Only moderate phenotypes could be associated with NDST3 deficiency, including small changes in high density lipoprotein and total cholesterol and altered anxiety related behavior. In mouse development, it appears to play no essential role, because all the major organ systems, including the brain, were unaffected morphologically, and no delay in development was obvious. This is in agreement with our finding that NDST3 deficiency did not result in dramatic differences in tissue HS composition. In this regard, NDST3 behaves much like NDST2, which when mutated also does not cause large changes in HS composition (32,33). The simplest interpretation of these findings is that NDST3, although being expressed, does either not contribute to HS sulfation in non-affected tissues, or that other Ndst isoforms may compensate for NDST3 deficiency in a tissue-dependent context. This idea is in agreement with the analysis of HS in different parts of the brain. Here, inactivation of NDST3 had highly variable effects, decreasing overall sulfation in the cortex dramatically while having little effect on brain stem or hippocampus. The former finding is reminiscent of HS sulfation in NDST1deficient mice, which show a general HS undersulfation (18,32), whereas the latter finding is reminiscent of the behavior of NDST2-deficient mice, which only show alterations in synthesis of highly sulfated heparin in connective tissue mast cells; other tissues were not affected (22,23,32). However, HS from the cerebellum showed an increase in N-and O-sulfation, which raises the enigmatic question of how inactivation of a sulfotransferase can cause an increase in overall sulfation. One model of HS biosynthesis suggests that several of the enzymes are present in a multienzyme complex termed the GAGosome (4). In this model, the GAGosome could vary in composition of specific Ndst isoforms dependent on their levels of expression or of other proteins that act as chaperones or scaffolding proteins in the system. Thus, one can imagine that in the cerebellar granule cells the GAGosome might preferentially contain NDST3, whereas in other parts of the brain other Ndst isozymes predominate. If the capacity to N-deacetylate and N-sulfate N-acetylglucosamine residues varies across the different isozymes, as has been shown (2), then altering the composition of the GAGosome could affect the composition of HS in an unpredictable way. Thus, substitution of NDST3 by other isozymes with greater capacity to sulfate the chain could explain the enhanced sulfation of HS observed in the cerebellum. Indeed, in situ hybridization showed strong overlapping expression of ndst2 in cerebellar granule cells, raising the possibility that NDST2 incorporation in granule cell GAGosomes results in HS oversulfation. Although this model is attractive, it is also possible that changes in expression of NDST3 affect other metabolic pathways, e.g. signaling reactions that then affect metabolism. Analysis of NDST1 Ϫ/Ϫ ;NDST3 Ϫ/Ϫ compound mutant embryos showed a strongly reduced overall sulfation level and more dramatic changes in HS structure than observed in NDST1 Ϫ/Ϫ animals, such as the complete loss of the UA2S-GlcNS disaccharide and strong reduction of the relative amount of the UA-GlcNS6S disaccharide. This indicates a partial ability of NDST1 to compensate for the loss of NDST3 and vice versa, and the potential of both isoforms to generate specific HS modifications. Consistent with this, doubly deficient embryos resembled NDST1 single mutants phenotypically (brain hypoplasia and facial dysmorphia (18,20)) but showed a higher frequency and severity of deficiencies (39% severe defects in the compound versus 14% in the NDST1 single mutant). Both findings thus suggest that NDST1 and NDST3 participate in the regulation of common pathways required for neural crest and forebrain development, e.g. FGF and HH signaling. The predominant role of NDST1 in mouse development is furthermore supported by the finding that NDST2 Ϫ/Ϫ ; NDST3 Ϫ/Ϫ mutant mice develop normally and are viable and fertile. The relative importance of the fourth member of the family, NDST4, awaits characterization of mutant mice lacking this isozyme.	2018-04-03T02:26:45.463Z	2008-06-13T00:00:00.000	{ "year": 2008, "sha1": "6a110fb2bc71842d07b9aa37c0764eccbceccf1e", "oa_license": "CCBY", "oa_url": "http://www.jbc.org/content/283/24/16885.full.pdf", "oa_status": "HYBRID", "pdf_src": "Highwire", "pdf_hash": "fd51c101d07ef6c2aaf4450a4a22fa0d6aaa5658", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
211065260	pes2o/s2orc	v3-fos-license	Thrombin-derived C-terminal fragments aggregate and scavenge bacteria and their proinflammatory products Thrombin-derived C-terminal peptides (TCPs), including a major 11-kDa fragment (TCP96), are produced through cleavage by human neutrophil elastase and aggregate lipopolysaccharide (LPS) and the Gram-negative bacterium Escherichia coli. However, the physiological roles of TCP96 in controlling bacterial infections and reducing LPS-induced inflammation are unclear. Here, using various biophysical methods, in silico molecular modeling, microbiological and cellular assays, and animal models, we examined the structural features and functional roles of recombinant TCP96 (rTCP96) in the aggregation of multiple bacteria and the Toll-like receptor (TLR) agonists they produce. We found that rTCP96 aggregates both Gram-negative and Gram-positive bacteria, including Staphylococcus aureus and Pseudomonas aeruginosa, and their cell-wall components LPS, lipid A, and lipoteichoic acid (LTA). The Gram-negative bacteria E. coli and P. aeruginosa were particularly sensitive to aggregation-induced bacterial permeabilization and killing. As a proof of concept, we show that rTCP96 reduces LPS-induced NF-κB activation in human monocytes, as well as in mouse models of LPS-induced subcutaneous inflammation. Moreover, in a mouse model of subcutaneous inoculation with P. aeruginosa, rTCP96 reduced bacterial levels. Together, these results link TCP-mediated aggregation of endotoxins and bacteria in vitro to attenuation of inflammation and bacterial levels in vivo. All wounds, whether caused by trauma, burns, or surgery, are at risk of becoming contaminated by bacteria, which could lead to infection. The ability to effectively counteract bacteria is of evolutionary significance to our survival. For this purpose, mul-tiple host defense systems have evolved, such as multiple host defense peptides (1)(2)(3)(4). Toll-like receptors (TLRs) 2 have an important role in the innate immune system by detecting a broad range of pathogen-associated molecular patterns (5). For example, TLR4 is activated by lipopolysaccharides (LPS) from Gram-negative bacteria, and TLR2 is activated by lipoteichoic acid (LTA) and peptidoglycan (PGN) from Gram-positive bacteria (6). The sensing of microbes and their products by TLRs is crucial in early responses to infection. However, an excessive TLR activation is deleterious, causing localized inflammation, such as that found in infected wounds, but also severe systemic responses, as seen in sepsis (7). Therefore, clearance and control of not only bacteria but various TLR activators, such as LPS and LTA, is critical to maintaining an effective antibacterial response while maintaining control of inflammatory responses. We have previously shown that proteolysis of thrombin by human neutrophil elastase (HNE) leads to the formation of thrombin-derived C-terminal peptides (TCPs) of roughly 2 kDa (8), which are present in wound fluids (9,10) and have been demonstrated to exert antiendotoxic functions in vitro and in vivo (8,(10)(11)(12)(13)(14). Apart from these smaller fragments, proteolysis of thrombin by HNE also generates 11-kDa TCPs that are present in wounds. Recently, we have shown that such TCPs can aggregate both LPS and Escherichia coli bacteria, leading to killing of the bacteria, and subsequent phagocytosis in in vitro models (15). This study builds upon and extends our previous work with the aim of understanding of TCPs spectrum of interactions with TLR agonists and bacteria, and importantly, the physiological role in vivo. Here, we demonstrate that a recombinant 96-amino acid TCP (rTCP96) aggregates both Gram-positive and Gram-negative bacteria and the products LPS, lipid A, and LTA. Finally, we show as a proof of principle, that such aggregation reduces LPS-induced inflammatory signaling as well as levels of Pseudomonas aeruginosa bacteria both in vitro as well as in vivo in experimental animal models. Antimicrobial effects of rTCP96 TCP96 represents an HNE-generated fragment, which is nine amino acids shorter (from the N terminus) than the B4 chain of ␥-thrombin ( Fig. 1) (15). We recombinantly expressed TCP96 (rTCP96) and evaluated its antimicrobial effect on the Gram-positive Staphylococcus aureus, Bacillus subtilis, and Enterococcus faecalis and the Gram-negative E. coli and two P. aeruginosa isolates (indicated as I and II in Fig. 2). The results showed that rTCP96 reduced the levels of particularly the Gram-negative strains E. coli and P. aeruginosa by 100 -1000fold, whereas the reduction of the Gram-positive S. aureus and E. faecalis was, albeit statistically significant, less marked (Fig. 2). To analyze whether the killing was mediated by bacterial permeabilization, we next employed live/dead staining, which uses propidium iodide (red color) to detect loss of membrane integrity. As seen, rTCP96 aggregated the bacteria, and killed (red) bacterial cells were observed in the aggregates of E. coli, P. aeruginosa, and S. aureus ( Fig. 3A and Fig. S1). The size distribution of the aggregates and the relative abundance of the respective size groups was recorded and is summarized in Fig. 3, B and C, respectively. The shorter TCP GKY25 killed the bacteria but did not cause aggregation (Fig. 3A, rightmost panels). We next selected E. coli and S. aureus bacteria for analysis using TEM after treatment with 5 M rTCP96. Membrane breaks and perturbations, compatible with the results using the live/dead assay ( Fig. 3A and Fig. S1) were observed (Fig. S2). Taken together, these results show that rTCP96 can induce aggregation and permeabilization of various Gram-negative and Gram-positive bacteria, leading to bacterial killing. Aggregation of rTCP96 in the presence of TLR ligands Next, we investigated the interaction between rTCP96 and various TLR ligands. We used Blue Native gels to determine the complex formations between rTCP96 (5 M) and LPS (E. coli, 0 to 500 g/ml) (Fig. 4A). Under the conditions used, rTCP96 alone is not detected in the gel, whereas increased amounts of LPS caused a significant increase in complexes migrating as ϳ700 kDa and larger (Fig. 4A). Furthermore, we measured a significant increase of ␤-sheet structure/aggregation by detecting thioflavin T1 (ThT) fluorescence in the samples of rTCP96 (5 M), which were treated with LPS (E. coli) in the concentration range from 10 to 500 g/ml in Tris buffer, pH 7.4 (Fig. 4B). Based on the results from the ThT experiment, for the next set of experiments, we chose to use 5 M rTCP96 and 50 g/ml of TLR ligands in Tris buffer, pH 7.4. We performed the ThT assay to determine the increase of ␤-sheet structure in rTCP96 treated with the TLR ligands LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), LTA (S. aureus), and PGN (S. aureus). We detected a significant increase in the ␤-sheet secondary structure of rTCP96, as reflected by an increase in ThT fluorescence in rTCP96 subjected to LPS (from E. coli and P. aeruginosa), lipid A (E. coli), and LTA (S. aureus). We did not observe any changes after addition of PGN (S. aureus) (Fig. 4C). Dynamic light scattering analyses corresponded with the results from the ThT assay and the electrophoresis results using Blue Native gel. A significant increase of hydrodynamic radius, compatible with the observed interactions between rTCP96 and the TLR ligands, was detected in the presence of LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), and LTA (S. aureus) but not in the presence of PGN (S. aureus) (Fig. 4D). Moreover, we employed microscale thermophoresis to measure the direct interaction between rTCP96 and TLR ligands in the solution (Fig. 4E). The K d (g/ml) constants for LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), LTA (S. aureus), and PGN (S. aureus) were 14 Ϯ 6, 16 Ϯ 5, 18 Ϯ 7, 15 Ϯ 6, and 1449 Ϯ 375, respectively, demonstrating that rTCP96 exhibited significantly lower binding affinity to PGN when compared with the other TLR ligands (Fig. 4F). Structural changes in rTCP96 triggered by TLR ligands Next, we used TEM to visualize amorphous aggregates of rTCP96 (of sizes from 0.5 to 5 m), which were formed after Thrombin fragments aggregate bacteria incubation with the different TLR ligands LPS, Lipid A, and LTA (Fig. 5A). rTCP96 did not aggregate in the presence of PGN. Correspondingly, an increase of ␤-sheet structure of rTCP96 was detected by CD measurements in the samples incubated with the two LPS forms, lipid A, and LTA (S. aureus). As above, no significant difference in the secondary structure of rTCP96 treated by PGN was observed (Fig. 5B). Inflammatory local environments can have a low pH (16,17). We therefore next investigated aggregation and structural changes of rTCP96 at pH 5.5. Using the ThT assay we found that LPS, lipid A, and LTA also induced ␤-sheet formation at low pH ( Fig. S3A). Correspondingly, CD analysis at pH 5.5 demonstrated changes in secondary structure of rTCP96 in the presence of the TLR ligands LPS, lipid A, and LTA (Fig. S3B). Molecular simulations of the TCP96-LPS and -LTA interaction CG simulations enabled us to study the large-scale aggregation propensity for TCP96 fragments in the presence of the lipid core components from different microbial products, including LPS from P. aeruginosa and E. coli, and LTA from S. aureus. The simulations in the presence of all molecules displayed an increase in aggregation over time, driven primarily by hydrophobic interactions (Fig. 6). Pairwise distance analysis supported visual analysis (Fig. 7), with LPS derived from different species forming stable co-aggregates in a 1:1 ratio. Notably, the 1:2 ratio of TCP96 with LTA lipids exhibited a greater propensity for aggregation compared with the 1:1 ratio, consistent with the smaller size of the lipidic component of LTA compared with that of LPS. The RDF for TCP96-TCP96 interactions was next calculated, as a measure of the mean variation in density as a function of distance from the protein fragments (Fig. S4). The RDFs indicate that TCP96s come into closer contact in the presence of LTA or LPS compared with when they are incubated alone. Based on the first peak in the RDF profiles, these differences in contact distances were shown to be significantly different, and also confirm that for each simulation the co-aggregation process had converged following the first 0.2 s of simulation (Table S1). Furthermore, the data lend additional support to the observation that a higher concentration of LTA is required compared with LPS for efficient TCP96 aggregation. Effects of rTCP96 on endotoxin response in monocytes We next used reporter THP-1 monocytes to detect effects on LPS signaling by rTCP96. rTCP96 significantly reduced the activation of NF-B/AP-1 triggered by E. coli LPS (Fig. 8A). The MTT viability assay did not show any significant cytotoxic effect of rTCP96 on THP-1 cells, which suggests that the reduction in the NF-B/AP-1 activation was due to the neutralizing effect of rTCP96 on LPS and not by any rTCP96-mediated toxic effects (Fig. 8A). Effects of rTCP96 on endotoxin and bacteria in vivo We next explored whether rTCP96 could suppress LPS-triggered local inflammation in vivo. For this, we utilized the (NF-B-RE-Luc)-Xen reporter mouse model and studied the effects of rTCP96 on subcutaneous inflammation induced by LPS. (NF-B-RE-Luc)-Xen reporter mice carry a transgene containing six NF-B-responsive elements and a modified firefly luciferase cDNA. The reporter gene is inducible by LPS and helps in in vivo studies of transcriptional regulation of the NF-B gene. LPS (25 g) was subcutaneously injected into the mice, either alone or with rTCP96 (25 g). The luminescent signal after subsequent addition of luciferin, reporting NF-B activation, was recorded using live bioimaging (IVIS Spectrum) (Fig. 8B). We detected a significant reduction of NF-B activation after 3 h in mice co-treated with rTCP96-LPS when compared with LPS treatment alone. rTCP96 alone did not yield any significant increase in NF-B activation (Fig. 8C). In the next experimental model, we wanted to simulate a situation of direct contamination with bacteria. Bioluminescent P. aeruginosa or S. aureus bacteria (10 6 cfu, cfu/animal) were incubated with buffer or rTCP96 and immediately injected subcutaneously in SKH1 mice. In this model, the bacterial dose used causes a transient, and self-limiting bacterial infection. The results showed that rTCP96 reduced the bacterial load of P. aeruginosa, as assessed by in vivo bioimaging (Fig. 8, D and E). However, no significant reduction of S. aureus was observed (Fig. 8, D and E). Discussion In this study, we present evidence that rTCP96 aggregates not only Gram-negative E. coli but also other Gram-negative bacteria, such as P. aeruginosa, as well as Gram-positive bacteria, such as S. aureus. Although both Gram-negative and Grampositive bacteria were aggregated in vitro, we observed that bacterial killing was most pronounced for the Gram-negative E. coli and P. aeruginosa in vitro. Correspondingly, rTCP96 only reduced P. aeruginosa in an experimental model of subcutaneous inoculation of P. aeruginosa, whereas no effects on S. aureus were observed in the in vivo model. From a clinical perspective, this observation corresponds with the well-established fact that E. coli and P. aeruginosa are less frequent as infective agents in acute surgical wounds. Notably, the majority of surgical site infections are caused by Gram-positive bacteria, of which one major agent is S. aureus (18,19). Although spec- A, separation on Blue Native gels followed by Western blot analysis shows an increase of higher molecular complexes of rTCP96 (5 M) with an increasing amount of LPS from E. coli (0 -500 g/ml). One representative image of four independent experiments is shown (n ϭ 4). B, rTCP96 was incubated with LPS at the indicated concentrations. ThT assay demonstrates a significant increase of ␤-sheet structure in the studied concentration range for LPS (n ϭ 6). C, ThT assay demonstrates aggregation of rTCP96 in the presence of LPS (from E. coli and P. aeruginosa), lipid A (from E. coli), LTA (from S. aureus) but not with PGN (from S. aureus) (n ϭ 6). D, dynamic light scattering analysis of the samples analyzed in C is presented. For B and C, , p Յ 0.05; , p Յ 0.01; ***, p Յ 0.0001. p values were determined using one-way ANOVA with Dunnett's multiple comparison test, and for D with two-way ANOVA with Sidak's multiple comparison test (n ϭ 6). Thrombin fragments aggregate bacteria ulative, these observations suggest that TCP-mediated aggregation at physiological conditions, such as in wounds, may preferentially target bacteria, such as E. coli and P. aeruginosa. We also show that TLR4 agonists such as LPS and lipid A from various bacteria (E. coli and P. aeruginosa) cause aggregation of rTCP96. Notably, the TLR2 agonist LTA from S. aureus Thrombin fragments aggregate bacteria exhibited a similar aggregating effect on rTCP96. On the other hand, PGN (from S. aureus), which is a TLR2 agonist, did not exert any aggregating effects on rTCP96. In this context, it is notable that a major chemical difference between PGN and the other ligands is the lack of an acyl tail component in PGN. Hence, the data suggest that the hydrophobic lipid tails of lipid A, LPS, and LTA seem to be crucial in mediating the aggregation with TCP96, and likewise in binding to TLR hydrophobic pockets. Of particular importance was that we observed an anti-inflammatory effect of rTCP96, as it significantly reduced LPS-induced TLR4 activation of human monocytes in vitro, indicating that confinement of LPS leads to reduced inflammatory signaling. Importantly, this in vitro observation was translated to the in vivo situation, showing that TCP96 can also reduce NF-B activation in the experimental model of endotoxin-induced inflammation in NF-B-RE-luc mice. Taken together, the present work on aggregation of TCPs induced by bacteria and their products, therefore, adds a further role to proteolyzed thrombin fragments in aggregation and amyloid formation for rapid confinement of endotoxins and microbes. This connection between host defense and aggrega-tion suggest that control of bacteria and its products is a common theme for many amyloidogenic proteins (20), such as ␤-amyloid-peptide variants (21,22), peptides from ␤-amyloid precursor protein, and the prion protein (13,23). Moreover, eosinophilic cationic protein, known to be released from eosinophils, aggregates and kills bacteria in vitro (24). It remains to be investigated whether these other proteins and peptides mediate similar activities as reported here for TCP96. Worldwide antimicrobial resistance is an increasingly serious threat to global public health that requires the urgent discovery of new therapeutic approaches (25)(26)(27). The activities of TCP96 on LPS delineate an endogenous mechanism by which aggregation-prone TCPs facilitate and control inflammation. From an evolutionary perspective, this activity is both logical and beneficial from an organism's point of view, illustrating that it is better to localize and attenuate inflammation than not to contain it (Fig. 9). This observation is of interest, as this suggests that potentially lethal systemic reactions, such as seen in endotoxin-induced shock, can potentially be avoided. Future in vivo work utilizing systemic models of endotoxin shock and bacterial infections with both Gram-negative as Gram-positive The panels indicate pairwise intermolecular distances between all eight TCP96 molecules in the simulation box, ranging from close contact (white) to more separated (black). Bright regions of the panels signify larger contact areas between molecules. Dark areas indicate that the respective molecules do not interact. Increased aggregation is apparent through the appearance of additional bright areas in the system containing LTA/LPS. Note that the bright area through the center of each matrix is due to self-contact. Ec, E. coli; Pa, P. aeruginosa; Sa, S. aureus. Thrombin fragments aggregate bacteria pathogens are mandated to explore the therapeutic potential of anti-infective concepts based on aggregation induced confinement. TLR ligands LPS from E. coli, LPS from P. aeruginosa, and lipid A from E. coli were purchased from Sigma-Aldrich. LTA and PGN from S. aureus were purchased from InvivoGen. Animals BALB/c tg(NF-B-RE-Luc)-Xen reporter male mice and SKH-1 hairless male mice, purchased from Taconic Biosciences, were used for all experiments. The animals were housed under standard conditions of light and temperature and had free access to standard laboratory chow and water. Purification of rTCP96 A bacterial expression system consisting of pET-15b plasmid in E. coli strain BL21 codon plus (DE3) RIPL (Invitrogen) was used to produce the rTCP96. We cultivated the bacteria in LB broth (Sigma-Aldrich) supplemented with 34 g/ml of chloramphenicol and 100 g/ml of carbomycin. Isopropyl 1-thio-␤-D-galactopyranoside (400 M; VWR), added at the mid-log phase, was used to induce peptide production in the bacterial system. The rTCP96 peptides were extracted and purified by immobilized metal affinity chromatography (nickel-nitrilotriacetic acid-agarose, Invitrogen) under denaturing conditions (8 M urea, 10 mM Tris, pH 7.4), extensively washed with 20 mM imidazole in 8 M urea, 10 mM Tris, pH 7.4, and then eluted by stepwise increasing concentrations of imidazole (200 mM). rTCP96 was desalted in 10 mM Tris, pH 7.4, by stepwise dialysis and concentrated using a 3-kDa molecular mass cut-off Amicon ultracentrifugal filter device (Millipore, Germany) and stored at 4°C prior to use. Peptide purity was confirmed via Tricine gel electrophoresis followed by Gel Code Blue Safe Protein staining (Thermo Scientific) and Western blotting. The protein concentration was determined by Nanodrop (ND 1000, Thermo Scientific) (15). Native gel analysis (BN-PAGE) We detected the interactions of rTCP96 and TLR ligands by BN-PAGE and Western blot analysis. rTCP96 (5 M) was incubated with TLR ligands for 30 min at 37°C. The samples were loaded under reducing conditions on BN-PAGE (Native PAGE BisTris Gels System 4 -16%, Invitrogen) according to the manufacturer's instructions, which we followed by Western blotting. Although the complex formation is known to be a process involving several intermediate steps, this assay is accepted in the field as an approximate measure of the binding capability between peptides and ligands. Viable count assay To determine the antibacterial activity of rTCP96, we used E. coli ATCC 25922, S. aureus ATCC 29213, P. aeruginosa ATCC 27853 (indicated as Pa I), P. aeruginosa ATCC 9027 (indicated as Pa II), B. subtilis ATCC 6633, and E. faecalis ATCC 29212. The bacteria were grown to mid-logarithmic phase in 5 ml of Todd-Hewitt (TH) medium. The bacteria were centrifuged, washed, and suspended in 5 ml of 10 mM Tris buffer, pH 7.4. Next, the bacteria (10 l, 3 ϫ 10 8 cfu; cfu/ml) were incubated with 5 M rTCP96, 5 M GKY25 (used as a positive control), or buffer control (10 mM Tris buffer, pH 7.4) for 2 h at 37°C. A dilution series of the incubated samples were plated on TH agar plates, incubated overnight at 37°C, and the cfu was calculated (15). Thrombin fragments aggregate bacteria Thioflavin T dye-binding assays Amyloid formation was determined using the dye ThT. Thioflavin T preferentially binds to the ␤-sheet structures of amyloidogenic proteins/peptides. For examination of the concentration dependence of the aggregation, we incubated rTCP96 (5 M) and LPS from E. coli (0, 10, 50, and 100 g/ml) in buffer (10 mM Tris, pH 7.4, and 10 mM MES, pH 5.5) for 30 min at 37°C. Moreover, rTCP96 was incubated for with 50 g/ml of each ligand (LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), LTA (S. aureus), and PGN (S. aureus)) for 30 min at 37°C before measurements. Two hundred microliters of the materials were incubated with 100 M ThT for 15 min in the dark (ThT stock was 1 mM stored in the dark at 4°C). We measured ThT fluorescence using a VICTOR3 Multilabel Plate Counter spectrofluorometer (PerkinElmer Life Sciences) at an excitation of 450 nm, with excitation and emission slit widths of 10 nm. The baseline (10 mM Tris, pH 7.4, or 10 mM MES, pH 5.5, buffer, LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), LTA (S. aureus), and PGN (S. aureus)) was subtracted from the signal of each sample (15,28). Circular dichroism spectroscopy We performed circular dichroism (CD) measurements on a Jasco J-810 spectropolarimeter (Jasco) equipped with a Jasco CDF-426S Peltier set to 25°C. The peptides were diluted to 5 M in buffer (Tris, 10 mM, pH 7.4, and MES, pH 5.5) and incubated with 10 -300 g/ml of LPS for 30 min, placed in a 10-mm quartz cuvette and, after extensive purging with nitrogen, scanned over the wavelength interval at 200 -260 nm (scan speed: 20 nm/min). We calculated the averages of five scans for each sample. For examination of time dependence, rTCP96 (5 M) was incubated for 10 and 120 min at 37°C in the absence or presence of TLR ligands (50 g/ml) in 10 mM Tris, pH 7.4. The baseline (10 mM Tris, pH 7.4, or 10 mM MES, pH 5.5, buffer, LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), LTA (S. aureus), and PGN (S. aureus)) was subtracted from the spectra of each sample (15,29). Transmission electron microscopy We visualized the aggregates formed by rTCP96 (5 M) in the presence of ligands (50 g/ml), such as LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), LTA (S. aureus), and PGN (S. aureus) during incubation for 30 min at 37°C. We examined 10 view fields (magnification ϫ4200) of the mounted samples on the grid (pitch 62 m) from three independent sample preparations using TEM (Jeol Jem 1230; Jeol, Japan) in combination with negative staining. The samples were adsorbed onto carbon-coated grids (copper mesh, 400) for 60 s and stained with 7 l of 2% uranyl acetate for 20 s. The grids were rendered hydrophilic via glow discharge at low air pressure (15). Fluorescence microscopy analysis of bacterial aggregates The viability of E. coli ATCC 25922, S. aureus ATCC 29213, and P. aeruginosa ATCC 27853 in the aggregates was assessed by using the LIVE/DEAD BacLight TM Bacterial Viability kit (Invitrogen, Molecular Probes, Carlsbad, CA). For this purpose, the bacterial suspension was prepared as described above for VCA. Bacteria were then treated by 5 M rTCP96 or 5 M GKY25 in 10 mM Tris, pH 7.4. The buffer was used for control. After 2 h, the samples were mixed with 3 l of the dye mixture for each ml of the bacterial suspension, as reported on the manufacturer's protocol, and incubated in the dark at room temperature for 15 min. At the end of incubation, 5 l of the stained bacterial suspension were trapped between a slide and an 18-mm square coverslip. We examined 10 view fields (1 ϫ 1 mm) of the mounted samples from three independent sample preparations using a Zeiss AxioScope A.1 fluorescence microscope (objectives: Zeiss EC Plan-Neofluar ϫ100/1.3 oil; camera: Zeiss AxioCam MRm; acquisition software: Zeiss Zen 2.6 (blue edition). For the analysis we measured the area of all aggregates possible to distinguish in each picture (43 for E. coli, 42 for P. aeruginosa, and 67 for S. aureus). Microscale thermophoresis We performed MST analysis using a NanoTemper Monolith NT.115 apparatus (Nano Temper Technologies, Germany). We used a Monolith NT Protein labeling kit RED-NHS (Nano Temper Technologies) to label 5 M rTCP96 according to the manufacturer's protocol. A constant amount of 2 M rTCP96 was mixed with increasing concentrations of LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), LTA (S. aureus), and PGN (S. aureus) in Tris buffer (10 mM, pH 7.4). Next, 10 l of each sample was loaded into standard glass capillaries (Monolith NT Capillaries, Nano Temper Technologies), and we performed the MST analysis (settings for the light-emitting diode and IR laser were 80%). K d constants were analyzed using MST software MO Affinity Analysis version 2.2.4 (15). Dynamic light scattering We performed dynamic light scattering (DLS) measurements to determine the hydrodynamic radii of rTCP96 with or without ligands: LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), LTA (S. aureus), and PGN (S. aureus). Wyatt QELS (Quasi-Elastic Light Scattering, Wyatt Technology Corporation) and Dawn EOS (enhance optical system, Wyatt Technology Corp.) equipped with a temperature-controlled microsampler instrument were used for DLS measurements. We incubated the samples for 30 min at 37°C under reducing conditions, and the scattered light was detected at 18 different angles simultaneously. Before the experiment, all samples were filtered through 0.22 m pore-sized microfilters (Sartorius, Germany). Aliquots of samples were manually loaded into the flow cell and measured at 37°C. All samples (1 M rTCP96 with or without ligands: LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), LTA (S. aureus), and PGN (S. aureus)) were measured at least 15 times. GKY25 peptide was used as a negative control and analyzed under the same conditions. For evaluation of time dependence of the aggregation, rTCP96 (1 M) was incubated for 10, 30, 60, and 120 min at 37°C in the absence or presence of LPS (E. coli), LPS (P. aeruginosa), lipid A (E. coli), LTA (S. aureus), and PGN (S. aureus) (5 g/ml) in 10 mM Tris, pH 7.4. The hydrodynamic radii were analyzed by Astra V software using Zimm modeling (15). Thrombin fragments aggregate bacteria Molecular dynamics simulations The coarse-grained (CG) models of TCP96 and E. coli LPS were taken from previously published work, with the closely related P. aeruginosa LPS model derived from that reported in Ref. 15. This involved switching the GL0 and GL5 beads in our original E. coli LPS CG model and shortening the lengths of the carbon lipid tails (30). The initial CG model for S. aureus LTA was initially constructed based on an atomic model kindly provided by Dr. T. J. Piggot. This model did not include the extended glycerol-phosphate units but was restricted to the hydrophobic diacylglycerol component and carbon lipid tails, which anchor LTA to the cell membrane, and are likely to be key in the aggregation of the molecule, as well as interacting with the hydrophobic TLR2 ligand-binding pocket (31). To study the aggregation behavior of TCP with different lipids, 5 ϫ 1-s simulation replicas were run with eight TCP96 fragments in the presence of: (i) eight S. aureus LTA molecules; (ii) eight P. aeruginosa LPS molecules; or (iii) eight E. coli LPS molecules. Also, 5 ϫ 1-s simulations of eight TCP96 fragments with 16 S. aureus LTA molecules were run, to check for possible side effects upon aggregation, given that the lipid component of LTA is significantly smaller than that of LPS. In all systems, the proteins and microbial products were randomly placed and solvated with water and neutralizing Na ϩ and Cl Ϫ ions. The simulations were run using the MARTINI force field (32) at 313 K and 1 bar, which was kept constant using the Berensden algorithm (33). All simulations and analysis were carried out using Gromacs 2018 (34), whereas rendering was performed using VMD (35). The pairwise distances between all the residues within the TCP molecules were determined and averaged. The radial distribution function (RDF) for TCP-TCP interactions was calculated for each simulation. For all analysis, each simulation was divided into blocks of 0.2 s, discarding the first block to account for equilibration. NF-B activity assay NF-B/AP-1 activation, in THP-1-XBlue-CD14 reporter monocytes, was determined after 20 -24 h of incubation according to the manufacturer's protocol (InvivoGen). Briefly, 1 ϫ 10 6 cells/ml in RPMI were seeded in 96-well-plates (180 l) and incubated with peptides (GKY25 1 M; rTCP96 0.5-0.05 M), LPS (10 ng/ml), or both overnight at 37°C, 5% CO 2 in a total volume of 200 l. The following day, the activation of NF-B/AP-1 was analyzed as the secretion of embryonic alkaline phosphatase. The supernatant (20 l) from the cells was transferred to 96-well-plates, and 180 l of Quanti-Blue was added. The plates were incubated for 2 h at 37°C, and the absorbance was measured at 600 nm in a VICTOR3 Multilabel Plate Counter spectrofluorometer. MTT viability assay Sterile-filtered MTT (Sigma-Aldrich) solution (5 mg/ml in PBS) was stored in dark at Ϫ20°C until usage. We added 20 l of MTT solution to the remaining overnight culture of THP-1-XBlue-CD14 reporter monocytes from the above NF-B activity assay in 96-well-plates, which were incubated at 37°C. The supernatant was then removed, and the blue formazan product generated in cells was dissolved by the addition of 100 l of 100% DMSO in each well. The plates were then gently shaken for 10 min at room temperature to dissolve the precipitates. The absorbance was measured at 550 nm in a VICTOR3 Multilabel Plate Counter spectrofluorometer. Mouse inflammation model BALB/c tg(NF-B-RE-Luc)-Xen reporter mice (Taconic, 10 -12 weeks old) were used to study the immunomodulatory effects of rTCP96 (25 g) after subcutaneous co-treatment with LPS (E. coli, 25 g). The samples were preincubated for 30 min at 37°C before injection. The dorsums of the mice (5 to 6 mice per treatment group) were carefully shaved and cleaned. Mice were anesthetized with isoflurane, and 200 l of the sample was injected subcutaneously. The animals were immediately transferred to individually ventilated cages and imaged 3 h later. We used an In Vivo Imaging System (IVIS Spectrum, PerkinElmer Life Sciences) for determination of NF-B activation, which plays a key role in the regulation of immune response during infection. Bioluminescence from the mice was detected and quantified using Living Image 4.0 Software (PerkinElmer Life Sciences). Fifteen minutes before the IVIS imaging, mice were intraperitoneally given 100 l of D-luciferin (150 mg/kg body weight) (14). Mouse model of subcutaneous infection Male SKH-1 hairless mice (12 weeks old), were anesthetized using a mixture of 2% isoflurane and oxygen. Overnight cultures of bioluminescent bacteria, P. aeruginosa Xen41 or S. aureus 229, were refreshed and grown to mid-logarithmic phase in TH media. Bacteria were washed for 15 min (5.6 ϫ 1000 rpm) and diluted with 10 mM Tris buffer, pH 7.4. rTCP96 (5 M) was then mixed with 10 6 cfu of the bacteria and incubated for 2 h at 37°C. A total of 100 l of mixture was injected subcutaneously into the mouse dorsum. In vivo bacterial infection was imaged by measuring bioluminescence in anesthetized mice using IVIS imaging and the data obtained were analyzed using Living Image 4.0 Software (PerkinElmer Life Sciences). 5 to 6 mice per treatment group were used (14). Ethics statement All animal experiments were performed according to the Swedish Animal Welfare Act SFS 1988:534 and were approved by the Animal Ethics Committee of Malmö/Lund, Sweden (permit numbers M88-91/14, M5934-19, and M8871-19). Animals were kept under standard conditions of light and temperature and water ad libitum. Statistical analysis The graphs of VCA, ThT, and DLS are presented as mean Ϯ S.D. from at least three independent experiments. We assessed differences in these assays using one-way ANOVA with Dunnett's multiple comparison tests and two-way ANOVA with Sidak's multiple comparison tests. All of the data were analyzed using GraphPad Prism 8 (GraphPad Software, Inc.).	2020-02-09T14:07:02.616Z	2020-02-07T00:00:00.000	{ "year": 2020, "sha1": "73586a198677a1d69469c9ea2c431625f46584bd", "oa_license": "CCBY", "oa_url": "http://www.jbc.org/content/295/11/3417.full.pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "627d456b8d11d4f8c8278f7f252ab9f9808a1325", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [ "Chemistry", "Medicine" ] }
2773026	pes2o/s2orc	v3-fos-license	Re-revisiting Andersen’s Behavioral Model of Health Services Use: a systematic review of studies from 1998–2011 Objective: This systematic review aims to assess the use and implementation of the Behavioral Model of Health Services Use developed by Ronald M. Andersen in recent studies explicity using this model. Methods: A systematic search was conducted using PubMed in April 2011. The search strategy aimed to identify all articles in which the Andersen model had been applied and which had been published between 1998 and March 2011 in English or German. The search yielded a total of 328 articles. Two researchers independently reviewed the retrieved articles for possible inclusion using a three-step selection process (1. title/author, 2. abstract, 3. full text) with pre-defined inclusion and exclusion criteria for each step. 16 studies met all of the inclusion criteria and were used for analysis. A data extraction form was developed to collect information from articles on 17 categories including author, title, population description, aim of the study, methodological approach, use of the Andersen model, applied model version, and main results. The data collected were collated into six main categories and are presented accordingly. Results: Andersen’s Behavioral Model (BM) has been used extensively in studies investigating the use of health services. The studies identified for this review showed that the model has been used in several areas of the health care system and in relation to very different diseases. The 1995 version of the BM was the version most frequently applied in the studies. However, the studies showed substantial differences in the variables used. The majority of the reviewed studies included age (N=15), marital status (N=13), gender/sex (N=12), education (N=11), and ethnicity (N=10) as predisposing factors and income/financial situation (N=10), health insurance (N=9), and having a usual source of care/family doctor (N=9) as enabling factors. As need factors, most of the studies included evaluated health status (N=13) and self-reported/perceived health (N=9) as well as a very wide variety of diseases. Although associations were found between the main factors examined in the studies and the utilization of health care, there was a lack of consistency in these findings. The context of the studies reviewed and the characteristics of the study populations seemed to have a strong impact on the existence, strength and direction of these associations. Conclusions: Although the frequently used BM was explicitly employed as the theoretical background for the reviewed studies, their operationalizations of the model revealed that only a small common set of variables was used and that there were huge variations in the way these variables were categorized, especially as it concerns predisposing and enabling factors. This may stem from the secondary data sets used in the majority of the studies, which limited the variables available for study. Primary studies are urgently needed to enrich our understanding of health care utilization and the complexity of the processes shown in the BM. and exclusion criteria for each step. 16 studies met all of the inclusion criteria and were used for analysis. A data extraction form was developed to collect information from articles on 17 categories including author, title, population description, aim of the study, methodological approach, use of the Andersen model, applied model version, and main results. The data collected were collated into six main categories and are presented accordingly. Results: Andersen's Behavioral Model (BM) has been used extensively in studies investigating the use of health services. The studies identified for this review showed that the model has been used in several areas of the health care system and in relation to very different diseases. The 1995 version of the BM was the version most frequently applied in the studies. However, the studies showed substantial differences in the variables used. The majority of the reviewed studies included age (N=15), marital status (N=13), gender/sex (N=12), education (N=11), and ethnicity (N=10) as predisposing factors and income/financial situation (N=10), health insurance (N=9), and having a usual source of care/family doctor (N=9) as enabling factors. As need factors, most of the studies included evaluated health status (N=13) and self-reported/perceived health (N=9) as well as a very wide variety of diseases. Although associations were found between the main factors examined in the studies and the utilization of health care, there was a lack of consistency in these findings. The context of the studies reviewed and the characteristics of the study populations seemed to have a strong impact on the existence, strength and direction of these associations. Conclusions: Although the frequently used BM was explicitly employed as the theoretical background for the reviewed studies, their operationalizations of the model revealed that only a small common set of variables was used and that there were huge variations in the way these variables were categorized, especially as it concerns predisposing and enabling factors. This may stem from the secondary data sets used in the majority of the studies, which limited the variables available for study. Primary studies are urgently needed to enrich our understanding of health care utilization and the complexity of the processes shown in the BM. Introduction Health care utilization is the point in health systems where patients' needs meet the professional system. It is well known that apart from need-related factors, health care utilization is also supply-induced and thus strongly dependent on the structures of the health care system. Furthermore, many study findings have shown differences in health care utilization based on patients' social characteristics. For instance, women tend to use outpatient health care services more often than men. In addition to the multitude of studies describing patterns of utilization in different health care settings, several scholars have developed explanatory frameworks identifying predictors of health care utilization [1]. One of the most widely acknowledged models is the Behavioral Model of Health Services Use (BM), which was developed in 1968 by the US medical sociologist and health services researcher Ronald M. Andersen [2], [3], [4], [5], [6], [7], [8], [9], [10] as a result of the third survey of the Center for Health Administration Studies and the National Opinion Research Center [5], [9]. The BM is a multilevel model that incorporates both individual and contextual determinants of health services use. In doing so, it "… divides the major components of contextual characteristics in the same way as individual characteristics have traditionally been divided-those that predispose …, enable …, or suggest need for individual use of health services" ( [9], p. 652). In their most recent explication of the model, Andersen & Davidson [3] described these three major components as follows: • Predisposing factors. Individual predisposing factors include the demographic characteristics of age and sex as "biological imperatives" ( [3], p. 7), social factors such as education, occupation, ethnicity and social relationships (e.g., family status), and mental factors in terms of health beliefs (e.g., attitudes, values, and knowledge related to health and health services). Contextual factors predisposing individuals to the use of health services include the demographic and social composition of communities, collective and organizational values, cultural norms and political perspectives. • Enabling factors. Financing and organizational factors are considered to serve as conditions enabling services utilization. Individual financing factors involve the income and wealth at an individual's disposal to pay for health services and the effective price of health care which is determined by the individual's health insurance status and cost-sharing requirements. Organiza-tional factors entail whether an individual has a regular source of care and the nature of that source. They also include means of transportation, travel time to and waiting time for health care. At the contextual level, financing encompasses the resources available within the community for health services, such as per capita community income, affluence, the rate of health insurance coverage, the relative price of goods and services, methods of compensating providers, and health care expenditures. Organization at this level refers to the amount, varieties, locations, structures and distribution of health services facilities and personnel. It also involves physician and hospital density, office hours, provider mix, quality management oversight, and outreach and education programs. Health policies also fall into the category of contextual enabling factors. • Need factors. At the individual level, Andersen and Davidson [3] differentiate between perceived need for health services (i.e., how people view and experience their own general health, functional state and illness symptoms) and evaluated need (i.e., professional assessments and objective measurements of patients' health status and need for medical care). At the contextual level, they make a distinction between environmental need characteristics and population health indices. Environmental need reflects the health-related conditions of the environment (e.g., occupational and traffic and crime-related injury and death rates). Population health indices are overall measures of community health, including epidemiological indicators of mortality, morbidity, and disability. The BM has frequently been used in studies, mainly those conducted in the United States and the United Kindom. It has also been applied in numerous systematic reviews on different aspects of health care utilization to structure their results [11], [12], [13], [14], [15], [16]. In other countries, such as Germany, only recently has there been increased awareness and use of the model. In Germany, for instance, it was adopted by the Federal Health Reporting System for analyzing health services utilization within the country [17], [18]. The present systematic review was conducted to assess the results of recent studies which explicitly employed the BM as their theoretical background. The broader rationale for this review was to explore: (1) the use of different versions of the BM, (2) the application and operationalization of the BM and (3) evidence for the influencing factors specified in the BM. Methods A systematic review was conducted between April and July 2011. Despite minor changes, the methods used for this review follow the PRISMA Statement (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [19]. The PRISMA statement, which provides a guideline for authors when reporting systematic reviews, contains a 27-item checklist and a four-phase flow diagram specifying important topics to be included in the Abstract, Introduction, Methods and Discussion sections. The present review, which aims to provide a theoreticallybased analysis of utilization in health care, was conducted as part of a larger scientific network project ("Utilization of health-related services in Germany -theoretical approaches, methods and empirical results in medical sociology," NWIN) funded by the Deutsche Forschungsgemeinschaft (German Research Foundation, grant no.: JA 1849/1-1). Literature search and study selection PubMed was selected as the only database for the systematic review and was searched in April 2011 (last search on April 24, 2011). The search strategy (see Table 1) aimed to identify articles in which the Andersen model had been applied and which had been published in English or German between 1998 and March 2011. The search was limited to studies conducted with adults in Europe and Anglo-American countries (USA, Canada, Australia and New Zealand) in order to achieve comparability. To ensure that all potentially relevant articles had been identified, the search terms used included Andersen's name as well as the terminology used in Andersen's Behavioral Model and derived models. No limits were set as to the study design used and whether the Behavioral Model had been compared to other approaches. Selection The selection of articles followed systematic review methodology and included the following steps: 1) title and author, 2) abtract and 3) full text. Inclusion and exclusion criteria were defined before each step (see Table 2). Each step of the selection process was conducted independently by two researchers (BB and DG). The results were then compared and, in cases of disagreement, discussed until a consensus was reached. Of the initial 328 studies retrieved through the PubMed search, only 16 met all of the inclusion criteria (see Figure 1). Data analysis In preparation for data synthesis, a data extraction form was developed to collect information on 17 categories: author, title, publication year, country, sample size, gender, age of participants, population description (e.g., ethnicity, social background), setting (e.g., outpatient or inpatient care), aim of the study, disease of interest, data source, methodological approach, variables researched, use of the Andersen model, applied model version, and main results. All of the studies were coded independently by both researchers based on these 17 pre-defined categories. All relevant information from the studies was collated into a table. An additional quality analysis (e.g., SIGN) was not conducted. Next, the original categories were further collated into the following six categories, which form the main basis for the presentation of the review's results: To present quantitative data (age, sample size, etc.), ranges were chosen as summary measures. Central themes and topics (e.g., methods, aims and settings) were extracted from the studies, summarized and presented according to their frequency of occurrence in the studies. The data collected from the studies was then further categorized based on the three main factors of the Andersen model (predisposing, enabling and need factors). Thus, all relevant topics were extracted from the studies. When new topics were identified, the category system was expanded. This information was then summarized for each category. Study characteristics The main study characteristics of the 16 studies included in this review are detailed in Table 3. The 16 studies included were published between 1999 and 2011 and were conducted in the US (n=9), Canada (n=4), US and Canada (n=1), Australia (n=1) or Germany (n=1). Secondary data, primarily from national surveys, were used as the sole source of information in almost all of the studies (n=14). In two studies, data were taken from hospital records [20], [21], and in the Australian study, data from Medicare records were used [22]. Primary data were only collected in three studies [22], [23], [24], one of which used focus groups in order to identify topics and develop questions for a subsequent survey [23]. The other two conducted a telephone survey [24] and a community survey [22]. All of the reviewed studies employed quantitative analysis methods. Multiple logistic regression analysis was the predominant statistical approach taken (N=13). Additionally, two studies applied linear regression models, and one study exclusively conducted a descriptive analysis. One study took a qualitative community-based participatory research (CBPR) approach with focus groups [23]. Seven studies examined health services utilization in general rather than concentrating on a specific health care sector. Two studies focused on outpatient care [25], [26], three focused on the primary care sector [18], [22], [27], one focused on physician and hospital services [28], one focused on tertiary care [20], and two focused on mental health services -in one case, services provided by a community center [21]; in the other, services by general practitioners and psychatrists [29]. Characteristics of the study populations Sample sizes The sample sizes of the studies varied from N=220 [24] to N=169,546 [30]. Gender Most of the studies (N=13) included both women and men. Two studies investigated only female immigrants from Latin America [21], [23] and one study exclusively investigated African American men [31]. Population description Nearly half of the studies (N=7) included samples of the general population. In six of these studies, the participants were recruited at the household or community level; in one study, they were recruited at an emergency department [20]. Three studies focused on low-or lowerincome populations [25], [26], [34]. Another three studies focused specifically on immigrants -two on Asians who immigrated to Canada [24], [29] and one on Latin American women who immigrated to the US [21]. Another three studies focused on specific ethnic groups. Two of these examined African Americans living in the US [31], [32] and one examined Latinas living in the US [23]. Use of the Andersen model Applied model versions Different versions of the Andersen model were found to have been used in the reviewed studies. In some studies, the authors indicated using more than one version of the model as the theoretical framework for their work. More than half of the studies [18], [20], [21], [22], [23], [26], [29], [30], [31], [33] applied the 1995 version of Andersen's Behavioral Model [2]. The Andersen-Newman Service Utilization Model (Andersen and Newman, 1973) was used in five studies [22], [24], [29], [30], [34]. Two studies [27], [32] employed Andersen and colleagues' Behavioral Model for Vulnerable Populations (Gelberg, Andersen, and Leake, 2000) and two studies [28], [29] applied Andersen's Behavioral Model of Families' Use of Health Services [5]. Two other studies [25], [29] used the Table 4 gives an overview of the key variables examined in the studies. It also indicates the number of studies that applied each variable. The most frequently researched predisposing factors were age, marital status, gender/sex, education, ethnicity/nativity and employment status. The enabling factors most often studied were income/financial situation, health insurance, having a usual source of care/family doctor, and availability of medical services/inpatient and outpatient care facilities. The most frequently examined need factors were health status (mental or physical), self-reported/perceived health, diabetes, depressive symptoms, hypertension, heart disease, cancer, number of prior medical/chronic conditions, and daily activities/limitation in daily activities. For the most part, the reviewed studies categorized variables in the same way. However, differences were found in the way some studies classified variables as predisposing and enabling variables. (Table 4 shows the variables as they were originally categorized in the reviewed studies.) In order to structure and present the review's findings in this article, the variables were pooled and assigned to only one of the categories (predisposing, enabling or need factors) based on Andersen & Davidson's 3 most recent explication of the model. In cases where recategorization was necessary, changes made to the original categorization have been noted. Predisposing factors Age Significant associations between age and utilization of health care services were found in the majority of the studies [18], [20], [21], [24], [25], [27], [28], [29], [30], [31], [32], [33], [34]. However, the direction of this association differed substantially depending on other participant characteristics. For example, it was found that the elderly were less likely to use alcohol, drug, and mental health services (ADM) than younger people [27]. Other studies found that older residents had low odds of specialist consultations [33], but high odds of four or more consultations with general practitioners [28], [31], [33]. In a sample of African Americans and Latinos living in the US, it was found that older participants were more likely to report having obtained a physician diagnosis for their medical conditions than their younger counterparts (70% of those ≥60 years vs. 58% of those ≤30 years) [32]. In young male and female immigrants, mental health consultations with general practitioners increased substan- [29], and in low-income Latina immigrants, older women were about 1.04 times as likely to use mental health services as their younger counterparts (p<.01) [21]. In Canada and the US, young women aged 18 to 44 were found to be more likely to contact a doctor than older women [28]. The same study found that women aged 45 to 64 had lower odds of being hospitalized than younger and older women and that men aged 45 to 64 had higher odds of being hospitalized than younger and older men [28]. Gender Associations were also frequently demonstrated between gender/sex and utilization of health care services [18], [22], [25], [30], [34]. Three studies analyzing gender/sex as a single variable reported that women were more likely to visit a physician than men [22], [30], [34]. These studies included samples of young Australian adults [22], US community-dwelling respondents [30] and low-income persons in the US [34]. Another study analyzing the variables age and gender/sex in combination found that younger age (19 to 39 years) and male gender were risk factors for not visiting a physician during the previous year [25]. Ethnicity/nativity Various studies reported associations between ethnicity or nativity and the utilization of health care services [25], [26], [27], [28], [29], [30], [33], [34]. Two studies conducted in the US found that black non-Hispanics, Hispanics, Asians or other racial ethnic groups [25], [30] and non-Hispanic others (e.g., Asians or Native Americans) [28] were significantly less likely than white non-Hispanics to receive treatment. Another US study of vulnerable subgroup members identified non-whites as having lower likelihoods of alcohol, drug, and mental health service use [27]. In Canada, visible minorities and Aboriginal people were found to be less likely to report specialist consultations than whites [33], and in the US, blacks were found to consume more physician care than other ethnic groups [34]. A sample of recent Chinese immigrants living in Canada differed in their mental health service utilization based on their geo-political origins. Whereas immigrants from Macau and Taiwan had the lowest rates of mental health visits to physicians, immigrants from mainland China and Hong Kong had higher rates (with the exception of young men from Hong Kong). The authors argue that these discrepancies may stem from ''sub-cultural'' differences among Chinese immigrants of different origins [29]. In the same sample, the length of time an immigrant had been living in Canada was associated with a significant increase in the rate of mental health visits in all sex and age groups [29]. Ethnicity, as a factor influencing health service utilization, was also looked at in combination with health insurance status. A study including a sample of lower-income adults in the US showed that among those insured, African Americans and American Indians/Alaska Natives were more likely than non-Latino whites to have a usual source of care. Furthermore, among the uninsured, having a usual source of care was more prevalent in Latinos than in non-Latino whites. However, the study found that Latinos, predominantly Mexican-origin Latinos, did worse (in terms of not having a usual source of care) in areas with a large low-income population [26]. Education In six studies, education was significantly associated with utilization of health care services [22], [25], [28], [29], [30], [31]. One study showed that African American men in the lowest education group had the lowest odds of scheduling a routine health examination [31]; another found that, among recent Chinese immigrants in Canada, men and women with university degrees were more likely to have mental health consultations than their less educated counterparts [29]. In one study, which considered education an enabling resource, Canadian and US adults with less income and education had fewer contacts with doctors than their counterparts [28]. In a sample of low-income children and adults residing in large metropolitan statistical areas in the US, those who were Latino and Asian with less education were more likely to have less access to medical care [25]. Similarly, an Australian study found that younger adults who visited a general practitioner were more likely to be educated [22]. Contrary to these findings, however, a random telephone survey in the US observed that respondents with less than a high school education were more likely to have used health professional services for the treatment of mental or emotional problems than those with at least some college education [30]. Marital status Marital status was repeatedly found to be associated with utilization of health care services [22], [23], [29], [30], [31], [32]. However, the direction of this association differed in the reviewed studies. One study examining routine health care services use among Latino women in the US reported that women who were not married were more likely to delay care than married women [23]. Similarly, an Australian study noted that married women were more likely to visit a general practitioner than unmarried women [22]. In a sample of African American men, those who were married were also more likely to schedule a routine health examination [31]. In the case of mental health services, it was found that persons who were previously married or never married were more likely to have received psychiatric treatment than those who were currently married [30]. A sample of recent Chinese immigrants in Canada revealed that unmarried women had lower rates of general practitioner visits for mental health problems than married women (RR=0.94 and 0.92); contrarily, single men had higher visit rates than married men (RR=1.22). Among studies examining multifactorial influences on health services utilization, one study found that medically under-served single African-American and Latino parents and those living with a smaller family more frequently reported having received a physician diagnosis [32]. Another study observed that younger Australian women who were separated, divorced, or living with children were more likely to visit a general practitioner compared to their counterparts [22]. Cultural norms and traditional health beliefs/behavior Cultural norms and traditional health beliefs or behavior were also identified in some studies as predisposing factors for the utilization of health care services. A community-based participatory study on Puerto Rican Latinas living in the Capital District of New York State identified women preferring to be seen by a Latino doctor (PRR 1.18) and those using alternative medicine (PRR 1.28) as being significantly more likely to delay care [23]. Traditional male role norms around the disclosure of vulnerability were found to have a negative impact on African American men's routine health examination receipt [31]. One study, which examined the impact of culture on the use of Western health services by older South Asian Canadians, found that having fewer cultural barriers and a lower level of agreement with South Asian health beliefs, but a stronger South Asian ethnic identity were related to using more types of Western health services [24]. This study originally considered cultural factors as a separate category. Trust in/familiarity with medical organizations Trust in and familiarity with medical organizations were also significantly associated with the use of health services. The odds of routine health examinations were lower in African American men who reported less trust in medical organizations and who believed that they should keep their concerns and emotions private [31]. Nonurgent patients who visited emergency departments without first seeking care from a primary care provider reported that they had done so because they felt more familiar with the emergency department (11%) and trusted its services (7%) [20]. Language Only one study explicitly reported associations between language and service use. The findings from this study suggest that language alone does not predispose individuals to the use of health services. The study, which investigated mental health care use among Chinese immigrants in Canada, found, for example, that whereas men who were able to speak English were more likely to visit a general practitioner, women who could speak English had a lower rate of psychiatric visits. The authors considered various reasons for this finding, including that immigrant men may feel greater pressure to succeed in their new country. In addition, they suspect that since Chinese immigrants in British Columbia receive health care primarily from Chinese-speaking general practitioners, their English skills may be less relevant to access to health care [29]. Region of residence (urban/rural) Region of residence was also associated with health care utilization. In one study, living in an urban area significantly increased the likelihood of residents to contact a doctor [18]. In another, residents of more rural counties or regions with fewer available health care services were less likely to visit a physician than their counterparts [34]. Community structure/social composition Two studies reported on the influence of community structure and social composition on health care utilization. One study found that lower-income adults living in a community with a large population that depends on the health care safety net were less likely to have a usual source of care [26]. In the other study, higher socioeconomic status of the neighborhood was associated with decreased utilization rates in young males [29]. Enabling factors Income/financial situation Associations identified between income and service use varied in the reviewed studies. US adults with lower income had lower likelihoods of doctor contacts [28] and less access to alcohol, drug, and mental health care [27]. Another US study, however, found that persons with annual household incomes less than $50,000 were more likely to receive psychiatric treatment than those with incomes of $75,000 or higher [30], and an Australian study with young adults found that those who reported financial problems during the previous year were more likely to visit a general practitioner [22]. In under-served African American and Latino populations, less financial strain was associated with physician diagnosis rather than selfdiagnosis [32]. Insurance Several studies found that being insured significantly increased the likelihood of service use or decreased the delay of health care in different population groups [23], [25], [26], [27], [28], [30], [34]. In the US, uninsured women from Latin America were significantly more likely to delay care than their insured counterparts [23]. A study on lower-income adults living in large urban communities in the US found that health insurance status influenced the relationship between the relative size of a community's safety-net dependent population and the access of lower-income adults [26]. Studies also found that service use varied by type of health insurance. According to one study, which investigated the influence of health sector market conditions on vulnerable subgroups' use of alcohol, drug, and mental health services, study participants with Medicare, Medicaid, private fully managed, and private partially managed insurance had a higher likelihood of utilization in areas with higher Health Maintenance Organizations (HMO) penetration than participants with other types of insurance or no insurance plan [27]. Another study found that poor Medicaid recipients with supplemental insurance were more likely to use care than those without supplemental coverage [34]. Usual source of care Having a usual source of care was repeatedly associated with increased service use [18], [28], [31], [34]. In a sample of African American men, those who had a usual source of care were found to be more likely to schedule and obtain a routine health examination [31]. Likewise, a study of adults in both Canada and the US showed that those who did not have a regular doctor were much less likely to have contacted a doctor over the previous year than those who did [28]. One German study even found that, contrary to the anticipated role of family doctors as gatekeepers who reduce the number of contacts with other doctors, people with a family doctor actually had more contacts with doctors overall [18]. Physician diagnosis/completion of a mental health intake evaluation In an under-served African American and Latino population, physician diagnosis was associated with better accessibility to medical services [32]. A study of low-income Latina immigrants found that completing a mental health intake evaluation predicted service use and that women who completed evaluations were 32.13 times more likely to use services than those who did not [21]. The study also discovered that longer elapsed time between referral to mental health services and completion of the evaluation significantly predicted less use of services [21]. Case manager The same study of low-income Latina immigrants found that using the services of a case manager significantly increased the likelihood of mental health services use. Women with case managers were 3.49 times more likely to use mental health services than those without case managers [21]. Availability of health-related information A study of under-served African Americans and Latinos in the US found an association between physician diagnosis of medical conditions -as opposed to self-diagnosis -and a greater availability of health-related information [32]. Affordability of medical care In the same study, affordability of medical care services was also associated with physician-based diagnosis of medical conditions [32]. Health care safety net supply For lower-income adults living in large urban communities in the US, the extent of a community's health care safety net, such as total Medicaid payments in the area and number of safety-net clinics, substantially influenced residents' realized access to services [26]. The same study also indicated that lower-income residents had better access to health services in metropolitan statistical areas where the Medicaid program paid providers more. The authors considered this an important form of direct support for safety-net providers [26]. Health care market Health care market conditions also influenced access to medical care. In one study of lower-income adults, greater competition among HMOs did not impact their access to medical care regardless of their insurance status [26]. Another study found that the existence of more federally funded community health centers and a less equal distribution of income positively influenced low-income adults' access to medical care [25]. Emotional social support Only one study, conducted in the US, reported associations between emotional support and use of health services. This study showed that people who always received emotional support were less likely to use health care services for the treatment of mental or emotional issues than those reporting to usually, sometimes, rarely, or never receive emotional support [30]. Need factors Evaluated health status Several studies reported significant associations between physical and mental status and the utilization of health services [18], [22], [23], [24], [28], [30], [33], [34]. In nearly all cases, poorer physical and mental health was a significant predictor of increased utilization. In one study, however, Canadian adults in fair or poor health were not significantly different from those in excellent health with respect to contacting a doctor [28]. In a US study of medically vulnerable population groups (the elderly, poor and uninsured), those with a history of diabetes, hypertension, or high blood cholesterol had significantly more physician visits than others [34]. Persons with more mentally and physically unhealthy days were found to be more likely to receive treatment than their counterparts [30]. Women with chronic diseases (PRR 1.24) were more likely to delay care than their counterparts [22], [23] and those using oral contraceptives were more likely to visit a general practitioner [22]. In Canadian and US adults, restriction of activity, chronic conditions, and depressive episodes had an increasing effect on doctor contacts and hospitalizations. Those who experienced any restriction of activity were more likely to use hospital and physician services. In the same sample, the likelihood of contacting a doctor was higher in persons with two or more chronic conditions (Canadian OR=3.64; US OR=3.85) compared to those with only one (Canadian OR=2.54; US OR=2.05). Additionally, Canadian and US adults who had experienced a depressive episode during the previous year were more likely to have been hospitalized than those who had not (Canadian OR=2.16; US OR=1.69) [28]. Perceived need and self-rated health Perceived need and self-rated health were also associated with health services use in some studies [20], [30], [34]. Respondents who rated their health status as poor or fair and experienced at least one disability per day [34] and those with less than excellent self-rated health were significantly more likely to receive treatment than those with excellent self-rated health [30]. A reason given by nonurgent patients for presenting to emergency departments rather than seeking care from primary care providers was perceived need (22%). Respondents believed that their respective condition was serious enough to warrant emergency care [20]. Discussion During the study selection process, it became apparent that the Behavioral Model had been applied to a broad range of health services sectors and diseases. Most of the initially identified studies had to be excluded as they focused on very specific health services sectors or diseases (e.g., long-term care, nursing homes, HIV, and dental care) and made drawing comparisons between studies nearly impossible. This, of course, limits the scope of the results analyzed in this systematic review. In addition, the vast majority of the studies reviewed were conducted in the US and Canada, which reduces the external validity of the review's findings. All of the included studies explicitly employed Andersen's Behavioral Model as their theoretical background. Most used the 1995 version of the model, which is the fourth revision of the model initially developed in the 1960s. This model version reflects the multiple influences on health services' use and on health status by including feedback loops that represent the mutual influence of outcome, predisposing factors, perceived need and health behavior [2]. Other model versions were also applied depending on the study subject and population (e.g., Andersen and colleagues' Behavioral Model for Vulnerable Populations [35] and Andersen's Behavioral Model of Families' Use of Health Services [5]). The most recent version of the model [3] was only used in one study, which is surprising given that the 2001 version is the most detailed explication of the model, and all but one of the reviewed studies were conducted after its publication. As evidenced by Table 4, although the studies investigated a wide variety of indicators, most of the studies focused on similar main variables, which we have labeled "key variables". The main predisposing variables examined were age, gender/sex, education, and ethnicity, and the main enabling factors were income/financial situation, health insurance, and having a usual source of care/family doctor. It is difficult to provide an overall assessment of the need factors examined in the studies since this review specifically focused on studies that used the Andersen model and it is possible that other studies on relevant variables (e.g., specific disease states as need factors) were overlooked. As previously mentioned, the majority of the reviewed studies conducted secondary data analyses, which means that the authors were forced to choose from among the variables collected in the original primary studies. The dominance of certain variables (i.e., key variables) may be the result of either a theoretical or a pragmatic decision. For the most part, the researched variables were categorized identically in the reviewed studies, with only some studies differing in their classification of predisposing and enabling variables. These differing classifications suggest that certain variables play a dual role in health services use, which corresponds to the 2001 version of the Behavioral Model. For example, the socioeconomic status of a neighborhood may be seen as both a predisposing factor (e.g., in terms of supply-induced demand effects) and an enabling factor due to its association with individual and community income. Also, age and sex may be categorized as individual demographic predisposing factors and, at the same time, used as proxies of need factors due to their associations with morbidity [29]. As evidenced by the findings presented in the Results section, it is nearly impossible to identify the factor having the "strongest influence" on health services utilization. Even for age, a seemingly simple indicator of service utilization, the findings showed inconsistencies in the strength and direction of this association. Since none of the studies used complex statistical methods, such as the testing of multivariate models, it is impossible to adequately assess the correlations between the examined variables. Therefore, the explanatory power of the results is restricted and is often limited to single indicators. The complexity of the Behavioral Model has clearly not been conveyed in the production and presentation of these results. There is a strong of need for (internationally conducted) primary studies that apply the Behavioral Model, that adequately operationalize the complexity of the model and that use complex statistical analysis that reflects its complexity (e.g., path analysis). Such studies would make it possible to go beyond examinations and descriptions of individual indicators and to better understand the associations between the main factors (predisposing, enabling, need) and investigate feedback loops.	2018-04-03T03:21:53.979Z	2012-10-25T00:00:00.000	{ "year": 2012, "sha1": "e5db1dfb4122d29484a37349a5ca5755b765d40b", "oa_license": "CCBYNCND", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "e5db1dfb4122d29484a37349a5ca5755b765d40b", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
198370013	pes2o/s2orc	v3-fos-license	Application of calixpyrrole modified silica for the removal of 4-chlorophenol from aqueous media 1 Laboratory of Analysis of Organic Compounds, Faculty of Sciences I, Lebanese University, Hadath, Beirut, 1500, Lebanon ismailabbas057@gmail.com (I.I.A.) (X.X) 2 Laboratory of Applied Chemistry, Faculty of Science III, Lebanese University, Tripoli, 1300, Lebanon bassemelhamaoui@gmail.com (B.E.H.) 3 Department of Chemistry, Faculty of Science, Beirut Arab University, Tripoli, 1300, Lebanon najmeddine.hilal@gmail.com (H.N.) Introduction Chlorophenols (CPs) are widely used as herbicides, pesticides, fungicides, wood preservatives and as intermediates in the production of pharmaceuticals and dyes [1,2]. CPs present in natural environments [3], and are environmental pollutants, discharged into the environment from many resources such as industrial, agricultural, and domestic activities [4][5][6] CPs in water have been classified as priority pollutants by US Environmental Protection Agency (EPA) [7] due to their toxicities and persistence within the environment [8]. Toxicity of CPs is proportional to the number of chlorine substituents on phenol ring [9]. Exposure to CPs in humans may affect muscles, nervous system, heart, liver and kidneys [10,11]. Thus the removal of CPs from wastewater has become a serious challenge to scientists. Different methods with varying degrees of success have been developed to remove CPs from wastewater. These include biological degradation, chemical oxidation, solvent extraction, adsorption, coagulation and liquid membrane permeation [12][13][14][15][16]. Disadvantages such as high cost, low selectivity and little efficiency are restricting the wide use of some of these treatment methods [17]. Adsorption is considered as a simple and effective method in removing inorganic pollutants from wastewater [18,19]. Many adsorbents such as clays, zeolites, resins and silicates have been investigated for the removal of phenol derivatives [20][21][22][23]. Mesoporous silicates are the most preferable adsorbents because they have a large surface area, high selectivity and large pore size [24,25]. The search for low cost and high selective adsorbents has led us to attach calixpyrrole derivative into silica backbone and to investigate the efficiency of the obtained material in removing CPs from aqueous media. Organic groups on the surface of silica create binding sites, which are affinitive to CPs. Calixpyrroles can interact efficiently and selectively with either cations, anions or neutral molecules depending on their pendant arms [26][27][28]. Therefore, the objective of this study was to evaluate the effectiveness of calixpyrrole modified silica for the removal of CPs from aqueous media. The effects of various operating parameters, such as pH, temperature, sorbent dosage and initial CPs concentration on the uptake capacity have been studied. Apparatus FTIR spectra were recorded with FTIR-6300, JASCO in wavenumber range of 400-4000 cm -1 using pressed KBr disks and a spectral resolution of 4.0 cm -1 . A KBr beam splitter and a sample bench purged with dry air. The FTIR spectra were recorded with pellets obtained by pressing a mixture of 1.0 mg of sample and 100 mg of KBr under reduced pressure. The morphologic characteristics of the modified silica materials before and after modification were determined by scanning electron microscopy (ASC-2100, SERON technology, Korea). The acceleration voltage is 20 kV and the work distance is about 25 mm. Thermogravimetric analysis was conducted by SETARAM-LABSYS Thermal Analyzer in the flow of N2 within 20-700 °C temperature range, with a heating rate of 3 °C/min. Absorbance spectra measurements were made using computerized double beam UV-Vis Spectrophotometer (Jasco V-530, Serial no A030161150) with wavelength range from 200 to 700 nm, spectral bandwidth 2 nm, time interval 1.0 nm, speed 1200 mm/min, and matched Quartz cell with length 10 mm. Preparation of modified silica The synthesis of calixpyrrole modified silica is shown in Figure 1. The two steps procedure is as follow: A mixture of 25 mL of 1.0 M SiCl4 and 10.0 g of activated silica in 150 mL freshly distilled dry dichloromethane (DCM) was stirred and refluxed under nitrogen gas with 3 mL trimethylamine (TEA) as a catalyst for 24 hrs. The bonded silica (II) was filtered and washed with DCM and toluene. 12.0 g of modified silica (II) was obtained. A mixture of 5.0 g meso-tetramethyl-tetrakis-[4-(2-ethoxy) ethoxyphenyl]calix [4]pyrrole [29], 10.0 g of modified silica (II) and 10 mL of TEA in 150 mL anhydrous toluene was refluxed under an inert atmosphere for 48 hours. The silica immobilized calixpyrrole (III) was filtered and washed in sequence with hot toluene, methanol, distilled water and methanol. Subsequently, the final product (III) was dried under vacuum at 120 °C for 24 hours and kept in desiccator over P4O10. Extraction of chlorophenols from aqueous media/batch method Batch experiments were carried out to investigate the ability of modified silica (III) to extract chlorophenols from aqueous media. Extraction was performed in 100 mL glass stoppered bottles at an adsorbent dose of 0.2 g, chlorophenol concentration of 450 mg/L, temperature of 298 K and contact time of 120 min. This was carried out by mixing 0.2 g of the modified silica (III) with 50 mL of chlorophenol solution. After equilibrium was attained, the solutions were centrifuged and the concentrations of chlorophenols remained in solutions were determined by UV-Vis spectrophotometry. This method was used to study the effect of modified silica (III) dosage, initial chlorophenol concentration, reaction time, solution pH, and temperature on the extraction process. The influence of pH was done at pH values of 2-12 with 0.1 M HCl or 0.1 M NaOH solutions. The effect of initial chlorophenol concentration was done by varying the chlorophenol concentration from 50 to 800 mg/L. Modified silica (III) dose was varied in the range 0.01-0.45 g to determine its effect. The influence of contact time was done by changing the contact time from 0.5 to 180 min. Temperature effect was studied by changing the extraction temperature from 288 to 308 K using a thermostat water bath for temperature control. Extraction of chlorophenols from aqueous media/column method In the column experiments, 2.0 g of modified silica (III) was packed in a Pyrex glass column (20×2.5 cm). Before use, distilled water was passed through the column to equilibrate and clean. Portions of 20 mL of chlorophenol solution (450 ppm) were fed in a down flow manner and effluents were collected at a flow rate of 5 mL/min. The concentration of chlorophenols in the eluent was analyzed by UV-Vis spectrophotometer after 30 min interval. The saturated modified silica was regenerated with 2.0 M HCl solution. This work was repeated many times until the silica material became saturated with chlorophenols. Characterization of silica materials FTIR analysis was mainly performed to confirm the attachment of calixpyrrole onto silica surface. The characteristic absorption bands of silica and silica immobilized calixpyrrole (III) are displayed in Figure 2. In the spectra of silica (I) and (II), the two characteristic peaks observed at around 801 and 460 cm -1 are corresponding to Si-O vibrations [30]. The band observed at 1076 cm -1 is assigned to Si-O-Si stretching vibration. The presence of adsorption water was reflected by ν(OH) vibrations [30] at around 3442 and 1662 cm -1 . The large and broad band near 3600 cm -1 is attributed to O-H stretching vibration. As far as the spectrum of silica (III) is concerned, new strong absorption bands are observed at 2672-2936 and 1419-1471 cm -1 in comparison with that of silica(II). These bands confirm the anchoring of calixpyrrole onto silica surface. The peaks at 1471 and 2936 cm -1 are characteristics of the benzene ring. The absorption band at 1471 cm −1 is associated with C=C stretching vibration of the benzene ring. Peaks at 2936 cm -1 are attributed to C-H stretches. The absorption band at 1419 cm −1 is attributable to C-N stretching vibration of the pyrrole ring. The band at 2388 cm −1 is related to N-H overtone. Morphology differences between silica (I) and modified silica (III) are provided by SEM images in Figure 3a-b. SEM micrographs illustrate that particles of silica (I) are relatively homogeneous. Figure 3b clearly illustrates the immobilization of calixpyrroles over the surface of silica particles. In order to get a profound insight into the microscopic structure, EDS elemental mapping is produced and the results are shown in Figure 4. The EDS data of silica immobilized calixpyrrole (III) confirmed the presence of carbon and nitrogen which were primarily absent in activated silica (I) and modified silica (II). This means that the anchoring of calixpyrrole onto silica surface was successful. Thermogravimetric curves of the activated silica gel (I) and silica immobilized calixpyrrole (III) are shown in Figures 5a-b. TG curves presented an initial mass loss of 6.52% which is attributed to the release of water physically adsorbed on the surface. The second loss of mass of 34.80% is due to the decomposition of the calixpyrrole immobilized on the surface. The last decomposition of 3.65% at 500 K is related to the condensation of the remaining silanol groups. Thermogravimetric analysis reflects the thermal stability of modified silica (III). The above results indicate that silica gel is successfully modified with calixpyrrole. Effect of mass of silica immobilized calixpyrrole (III) Different masses (0.01-0.45 g) of calixpyrrole modified silica (III) were added into 200 ppm chlorophenol solutions in 50 mL volumetric flasks. The mixtures were mechanically shaken for 30 minutes to facilitate the interaction of chlorophenol with modified silica (III). The quantity of chlorophenol removed (% E) from the aqueous media was calculated by using the Equation (1). where Ci and Ce represent the initial and equilibrium concentrations of chlorophenol in the aqueous solution. The results are shown in Figure 6. It is evident that the amount of chlorophenol removed from solution increases as the mass of modified silica increases for a given initial concentration. This increase may be attributed to the existence of more binding sites (OH and NH units) for interaction with chlorophenols. The highest quantity of chlorophenol (95.6 %) was extracted by using 0.20 g of modified silica (III). Percent extraction was not changed significantly when the mass of extractant was larger than 0.20 g. So this mass (0.20 g) was used in studying the effects of other parameters (time, temperature, pH and concentration). Uptake capacity Uptake capacity is an important parameter to investigate because it reflects the efficiency of extractant to quantitatively remove contaminants from aqueous media. 0.20 g of modified silica (III) was equilibrated with a series of different concentrations (50-800 ppm) of chlorophenol solutions and the batch method was applied. Uptake capacity (qe) was calculated according to the Equation (2). [31] 300 Activated carbon [32] 136.2 Amberlite XAD-16 [33] 291.6 Granular Activated Carbon [34] 40 Porous magnetic resin-g-chitosan beads [35] 97 Multiple walled carbon nanotubes (MWNT) [36] 133 where V is the solvent volume (L), m is the weight of extractant (g), and Ci and Ce are the initial and equilibrium concentrations of chlorophenol in the aqueous solution, respectively. It was observed that the uptake capacity of modified silica (III) increases as the initial concentration of chlorophenol increases till it reaches maximum at 450 ppm ( Figure 7). This trend indicates that the binding sites of modified silica (III) are incapable to interact any further with free chlorophenol molecules in solution. The maximum uptake capacity of modified silica (III) is 229 mg/g. It is worth mentioning that chlorophenols were not extracted with activated silica (I) and modified silica (II). The results show that modified silica (III) has a high uptake capacity for chlorophenol which is bigger or similar to the other materials reported in the literature (Table 1). Effect of pH A series of 450 ppm chlorophenol solutions were transferred into glass stoppered bottles and the pH was adjusted to the desired value (2-12) with 0.1 M HCl or 0.1 M NaOH solutions. 0.20 g of modified silica (III) was added into each bottle and the mixtures were stirred for 30 minutes. An inspection of Figure 8 indicates that the removal of chlorophenols by modified silica (III) slightly increased when the pH of solution increased from 2.0 to 10.0. Further increase in pH of the media shows a minor decrease in the percentage of removal of chlorophenols. This may be attributed to the competition between hydroxide ions and chlorophenols. Effect of temperature Extraction experiments are conducted at various temperatures in the range of 288 to 308 K. The uptake capacities of modified silica (III) for the chlorophenols at different temperatures are shown in Figure 9. The results show that the capacity for chlorophenols rises by increasing the temperature of the solution. This method was also used to study the thermodynamics of the extraction process. Gibbs free energy (ΔG°) was calculated using the Equation (3). ΔG° = − RT ln Kd where, R is the universal gas constant (8.314 J/mol.K), T is the temperature (K) and Kd is defined by the equation: where Ca is the chlorophenol ion concentration removed from solution at equilibrium and Ce is the equilibrium chlorophenol concentration in solution. Furthermore, the enthalpy (ΔH°) and entropy (ΔS°) parameters were calculated from the slope and intercept of the plot of ln Kd vs. 1/T respectively by using Van't Hoff equation [37]: The results obtained are presented in Figure 10 and Table 2. The positive value of standard enthalpy (6.375 kJ/mol) indicates that the uptake of chlorophenol by modified silica (III) is an endothermic process which explains the fact that uptake efficiency increases with increase in temperature. The positive standard entropy change (23.23 J/mol.K) suggests the increase in randomness at the solid-liquid interface. Effect of contact time and uptake kinetics Modified silica (III) (0.20 g) were equilibrated with 50 mL samples of 450 ppm chlorophenol solution in glass-stoppered bottles for a fixed period of time (0.5-60 minutes). It can be observed that the rate of uptake of chlorophenols from aqueous media is fast at the beginning (first 5 minutes) and then diminished gradually with time until equilibrium is reached after 15 minutes (Figure 11). The concentration difference between the chlorophenol solution and the surface of modified silica (III) accelerates the diffusion process of these molecules from the liquid-phase to solid, causing the fast increase of uptake in initial stage [38]. The initial fast uptake rate may be also attributed to the availability of accessible binding sites on the surface of modified silica (III). The slow uptake in the later stages is probably due to the less availability of binding sites for interaction with chlorophenols. The mechanism of chlorophenol uptake by modified silica (III) was analyzed by fitting experimental data to the pseudofirst-order (Equation 6) [39], and the pseudo-second-order (Equation 7) [40] kinetic models. where qe and qt are the amounts of chlorophenol removed (mg/g) at equilibrium and at time t (min), respectively. 11. Effect of contact time on chlorophenol removal from aqueous media by modified silica (III) at 298 K. k1 (1/min) is the pseudo-first-order rate constant and k2 (g/mg min) is the pseudo-second-order rate constant; The values of k1 and qe were calculated from the slope and intercept of the plot of ln (qe−qt) vs. t, respectively. Likewise, the values of qe and k2 were calculated from the slope and intercept of the plot of (t/qt) vs. t, respectively. The experimental value of qe obtained from pseudo-firstorder kinetic (Figure 12) is different from the calculated qe value. In contrast, pseudo-second-order plot fitted data very well ( Figure 13). The constants of pseudo first-order and pseudo second order models are given in Table 3. The correlation coefficients (R 2 ) indicate that the uptake processes could be well defined by pseudo-second-order model under the experimental conditions. In addition, there is a high agreement between the calculated and experimental capacity values obtained from the pseudo-second-order kinetic model. This shows that chemical interaction is the rate-controlling step [41]. This also suggests that the number of binding sites on the silica surface and the number of chlorophenol molecules in the liquid phase together determine the rate of uptake [42]. Column studies Modified silica (III) was practically tested via column uptake studies. The plot Ct/Co vs t produced the breakthrough curve ( Figure 14). The breakthrough time and exhaustion time was found to be 5 and 14 hr, respectively. The maximum column capacity, qtotal (mg/g) for a given flow rate and inlet concentration is equal to the area under the plot of the adsorbed chlorophenol concentration, Cad (mg/L) vs time (t, min) and is determined from Equation 8 [43]: where Q, A and ttotal, are the volumetric flow rate (ml/min), the area under the breakthrough curve and the total flow time (min) respectively. The capacity of the modified silica (III) column for chlorophenol was found to be 7.8 mg/g. The uptake capacity of removal of chlorophenols by column technique is lower than that obtained by batch uptake studies. This may be attributed to the fact that the extraction of chlorophenols by column technique depends on several factors such as the flow rate of the mobile phase and the type and amount of mobile and stationary phases [28]. In this study, three models (Thomas, Yoon and Nelson and Yan et al.) were applied to describe kinetic studies. The linearized form of Thomas mode [44] is expressed as follows: where qT is the maximum capacity of uptake (mg/g), C0 and Ce are the inlet and outlet concentrations (mg/L) of chlorophenol respectively, Q is the volumetric flow rate (mL/min), Kt is the Thomas rate constant (mL/g.h), and M is the total mass of adsorbent. kT (0.932 mL/mg.hr) and qT (6.41 mg/g) were calculated from the slope and intercept of linear plot of ln[(Co/Ce)-1] vs t at a given flow rate ( Figure 15). The results are given in Table 4. Thomas model indicates that the rate driving force obeys 2 nd order reversible reaction kinetics. The linearized form of Yoon-Nelson model is represented as follows: where Ce and C0 are the effluent and influent concentrations, τ is the time required for 50% adsorbate breakthrough (min), kYN is Yoon-Nelson constant and t is the time (min). A plot of ln[Ce/(Co-Ce)] versus t provides a straight line with a slope of kYN, and intercept of -τ kYN ( Figure 16). The results are listed in where qy is the maximum uptake capacity (mg/g) of adsorbent and ky is the kinetic rate constant (mL/min.mg). ky and qy were calculated from the slope and intercept of the plot of ln[Ce/(C0-Ce)] vs. ln t ( Figure 17). Both Thomas (R 2 =0.993) and Yoon-Nelson (R 2 = 0.994) models show better fitting than Yan's model (R 2 =0.985). The highest sorbent capacity given by Thomas model is 6.41 mg/g which is consistent with that's calculated by integrated area method. The highest breakthrough time for 50% removal estimated by Yoon-Nelson model is around 6.68 h (Table 4). A useful extractant in wastewater treatment should have high uptake capacity for toxic pollutants and a good regeneration property. Regeneration of utilized modified silica ((III) was done with 2.0 M HCl to desorb the attached chlorophenols followed by washing with distilled water [28]. Recycled silica (III) was used for consecutive five cycled for the removal of chlorophenols ( Figure 14). The integration method was applied to calculate the maximum uptake capacity after first regeneration and it is determined to be 7.3 mg/g. This implies that modified silica (III) may be suitable adsorbent for chlorophenol removal from wastewater. The uptake capacity of regenerated modified silica (III) was not significantly reduced in the first four cycles. Then in the fifth cycle, capacity significantly decreased (0.32 mg/g), the breakthrough curve became steeper and the breakthrough happened faster. Conclusion Based on the results obtained in this study, calixpyrrole modified silica(III) was confirmed as an effective extractant for 4-chlorophenols. The removal efficiency was dependent on experimental parameters such as the amount of extractant, contact time, pH, initial concentration of chlorophenol and temperature. This material showed high uptake capacity (229 mg/g) for chlorophenol in aqueous solution which suggests the possibility of using this material in wastewater treatment. Thermodynamic studies indicated that the extraction process is endothermic and has a spontaneous nature. The uptake rate for chlorophenol was obtained to be conforming to the pseudo-second order kinetic model. The chemical stability of the modified silica (III) in acidic media, as well as the regeneration by washing with 2.0 M HCl, allowed the reuse of this material for many cycles.	2019-07-26T08:41:34.755Z	2019-06-30T00:00:00.000	{ "year": 2019, "sha1": "310587aca42284234d0ad3f732f7c47872ca6c60", "oa_license": "CCBYNC", "oa_url": "http://www.eurjchem.com/index.php/eurjchem/article/download/1846/pdf_1846", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "310587aca42284234d0ad3f732f7c47872ca6c60", "s2fieldsofstudy": [ "Environmental Science", "Chemistry", "Materials Science" ], "extfieldsofstudy": [ "Chemistry" ] }
199128543	pes2o/s2orc	v3-fos-license	Investigation of Interaction Solutions for Modified Korteweg-de Vries Equation by Consistent Riccati Expansion Method A consistent Riccati expansion (CRE) method is proposed for obtaining interaction solutions to the modified Korteweg-de Vries (mKdV) equation. Using the CRE method, it is shown that interaction solutions such as the soliton-tangent (or soliton-cotangent) wave cannot be constructed for the mKdV equation. More importantly, exact soliton-cnoidal periodic wave interaction solutions are presented. While soliton-cnoidal interaction solutions were found to degenerate to special resonant soliton solutions for the values of modulus (n) closer to one (upper bound of modulus) in the Jacobi elliptic function, a normal kink-shaped soliton was observed for values of n closer to zero (lower bound). Introduction The direct study of exact solutions to nonlinear evolution equations (NLEEs) has received much attention from many mathematicians and physicists due to the fact that new strides in nonlinear science, which were made possible by a substantial increase in computational platforms such as Mathematica, Maple, and MATLAB, have enabled improvements in the performance of complicated and tedious numerical computational methods. Indeed, several powerful methods such as the Tanh-function method [1][2][3], F-expansion method [4], Jacobian elliptic function method [5], and variational approach [6,7] have been proposed for constructing exact solutions to NLEEs. Despite the successful implementation of such methods, it is still challenging to obtain solutions for interactions among different types of nonlinear excitations such as the soliton-soliton interaction. Recently, some new soliton structure solutions were obtained for nonlinear systems. Chen et al. studied the vortex solitons in Bose-Einstein condensates with spin-orbit coupling and Gaussian optical lattices, based on the analytical and numerical method [8]. Milan et al. found exact fundamental soliton solutions in the spiraling guiding structures by the modified Petviashvilis iteration method [9]. Cheng et al. investigated the formation and propagation of a multipole soliton in a cold atomic gas with a parity-time symmetric potential using the modified square operator method [10]. Liu et al. obtained the three-soliton solutions for high-order nonlinear Schrodinger equation by Hirotas bilinear method [11]. Specially, Lou [12] proposed a consistent Riccati expansion (CRE) method, which is a more generalized yet simpler method to find interaction solutions for various NLEEs [13][14][15][16][17]. The core concept of CRE is the construction of interaction solutions based on the usual Riccati equation method and the consistent equation or the -equation [12]. The CRE method is critical to finding more new solutions to the -equation. In this study, the CRE method is used to construct several types of interaction solutions for the focusing real modified Korteweg-de Vries (mKdV) equation [18] shown in where and are arbitrary constants. The mKdV equation plays an important role in describing some physical phenomena, such as optical cycles [19,20], soliton propagation in plasma [21] and lattices [22], the Schottky barrier transmission lines [23], and fluid mechanics [24]. To provide better insights into these physical phenomena, finding and analyzing exact solutions to the mKdV equation is important. Previously, many powerful methods have been proposed for constructing exact solutions to the mKdV equation. For instance, in 1972, Hirota obtained an exact solution to the mKdV equation for the case of multiple collisions of solitons with different amplitudes [25]. Subsequently, he also derived the exact envelope soliton solution to the mKdV equation [26]. In 1973, Ablowitz et al. obtained exact solutions to the mKdV equation by using the inverse scattering technique [27]. In 1988, Akhmediev et al. used the Darboux transformation scheme to obtain second-order periodic solutions to the mKdV equation [28]. In 2004, Kevrekidis et al. derived some classes of periodic solutions to the mKdV equation by using direct methods [29]. In 2015, Jiao and Lou constructed a new soliton-cnoidal periodic wave interaction solution by using the CRE method [30]. However, they did not investigate how the soliton-cnoidal interaction solutions may be used to derive soliton-soliton or solitonperiodic wave interaction solutions among other types of solutions. Moreover, new interaction solutions to the mKdV equation involving different types of nonlinear waves must be investigated in depth. The present article is structured as follows. Section 2 introduces the CRE solvability of the mKdV equation. Section 3 describes new explicit interaction solutions such as soliton-soliton, multiple resonant soliton, soliton-cosine wave, and soliton-cnoidal wave solutions to the mKdV equation obtained using the CRE method. Furthermore, it is demonstrated that interaction solutions such as the soliton-tangent wave solution cannot be constructed for the mKdV equation. The last section presents a summary and discussion. CRE Solvability of the mKdV Equation Consider the following NLEE, shown in (2), with independent variables ≡ ( , 1 , 2 , . . . , ) and a dependent variable ≡ ( ) ( , , , where is a polynomial function of some arguments with the subscripts denoting partial derivatives. We assume that the solution to (2) is the following possible truncated expansion form where is determined from the leading order analysis of (2). All the expansion coefficient functions ( ) are determined by substituting (3) into (2) and then vanishing all the coefficients for a given power of ( ). Further, and are functions of ( , , ) and ( ) satisfies the following simple Riccati equation shown below: which includes the following five special solutions [31]. For < 0, For > 0, Definition. If the equation for ( = 0, 1, . . . , ) and , obtained by vanishing all the coefficients of each power in ( ) after the substitution of (3) into (2), is either consistent or not overdetermined, then the expansion in (3) is considered a CRE and the nonlinear system in (2) is said to be CRE solvable [8]. According to the CRE method defined above, one can obtain the following form based on the leading order analysis of the mKdV equation in (1) where 0 , 1 , and are functions of ( , , ) and ( ) satisfies the Riccati equation (see (4) above). Substituting (8) and (4) into (1) and vanishing all the coefficients of different powers of ( ), one obtains Based on the definition above, (11) is the consistent equation of the mKdV equation (or the mKdV -equation). If is a solution to the MDWW -equation in (11), the mKdV equation in (1) is CRE solvable. In this study, we set = −6 and = 1. Thus, the solutions to the mKdV equation are expressed as follows. Mathematical Problems in Engineering 3 Interaction Solutions to the mKdV Equation Upon the determination of solutions to (11) by using (12), the corresponding solutions to the mKdV equation in (1) can be obtained. In this section, we construct interaction solutions to the mKdV equation by using different types of trivial solutions to (11). . . Soliton-Soliton Interaction Solutions to the mKdV Equation. To obtain soliton-soliton interaction solutions to the mKdV equation, we consider the following form in (13) as the trial solution to (11): where ( ) satisfies the following Riccati equation in where is an arbitrary constant. This equation has special solutions similar to those in (5a), (5b), (6a), (6b), and (7). By vanishing all the coefficients for each power of after the substitution of (13) and (14) into the mKdV -equation in (11), one can obtain From (15), it can be seen that both and r are less than 0 when 1 ̸ = 0. Based on (5a), (5b), (6a), and (6b), Eqs. (4) and (14) have only solitary solutions (viz., (5a) and (5b)) but not tangent or cotangent solutions such as the ones described in (6a) and (6b). This shows that interaction solutions such as soliton-tangent (or soliton-cotangent) wave cannot be constructed for the mKdV equation. . . Interaction Solutions between Soliton and Cnoidal Wave for the mKdV Equation. In [26], Jiao and Lou constructed a solution of the following form for (11) where sn( , ) is the usual Jacobi elliptic sine function and ( , ], ) is the third type of incomplete elliptic integral. Jiao and Lou used the following parameters in (21) to obtain a special soliton-cnoidal wave interaction solution to the mKdV equation: As seen from (21), Jiao and Lou chose the modulus ( 1 ) of Jacobi elliptic function to be 1.5, which is outside the allowed range (0 < 1 < 1) [13]. In this study, we will further investigate how solitoncnoidal interaction solutions can be used to derive solitonsoliton and soliton-periodic wave interaction solutions among other types of solutions. To this end, we performed all the substitutions and evaluations by using the Mathematica software. Consider a trial solution of the following form for solving (11) where satisfies the following elliptic equation: Substituting (22) and (24) From the analysis of (24), we assume the solution of (24) in the following form Substituting (26) into (24) and setting the coefficient of {sn( ( 2 + 2 ), ), cn ( ( 2 + 2 ), ), dn( ( 2 + 2 ), )} equal to zero, one obtains Based on (25) and (27), one can find a group solution 2 ( 1 + 2 0 + 2 1 ) To investigate how the soliton-cnoidal interaction solutions could be used to derive soliton-soliton interaction or other types of solutions, we illustrate the following two cases corresponding to the soliton-cnoidal wave interaction solution described in (29) by selecting different sets of parameters. For the first case, the parameters are chosen as = −1, While Figure 1 shows two-dimensional views for interaction solution at = 0 and = 0. Figure 2 displays threedimensional plots for the evolution of soliton-cnoidal wave interaction solution with different values for the modulus in the Jacobian elliptic function, viz., = 0.000001, 0.5, and 0.99999. While = 0.5 exhibited a particular periodic-kink soliton wave interaction, the extreme values of = 0.00001 (a value close to the lower modulus limit or 0) showed a normal kink-shaped soliton and = 0.9999 (a value close to the upper modulus limit or 1) displayed an interaction between a periodic wave and another periodic wave. For the second case, the parameters were altered as shown in (32) Similar to the first case, we illustrate the structures of the soliton-cnoidal wave interaction solution for different values of = 0.00001, 0.5, and 0.99999. Clearly, as shown in Figures 3 and 4, wavenumbers and the amplitudes in the range of x(-40,40) and t (-40,40) are less than that of the first case (cf. Figures 1 and 2 ). While there is still a normal kink soliton in the x-u plot for n = 0.00001, an incomplete kink soliton is observed in the t-u plot in contrast to the first case shown in Figure 1(b). Building on the above two cases, soliton and soliton-soliton wave interaction solutions are derived from the soliton-cnoidal wave interaction solution by making the limit of the modulus approach either 0 or 1. Summary and Discussion In this study, we investigated the focusing mKdV equation by using the CRE method. This nonlinear equation was shown to be CRE solvable and interaction solutions; namely, solitonsoliton, soliton-trigonometric periodic waves, and solitoncnoidal periodic wave for the mKdV equation were explicitly provided by choosing different trial solutions for the mKdV w-equation shown in (11). In addition, analytical solutions for interactions between soliton and cnoidal wave were provided and their properties were discussed graphically. According to the presented analysis, soliton and soliton-soliton wave interaction solutions can be derived from the soliton-cnoidal wave interaction solution by making the limit of the modulus approach either 0 or 1 (i.e., lower or upper bounds for the modulus in the Jacobi elliptical function). Data Availability No data were used to support this study. Mathematical Problems in Engineering 7 Conflicts of Interest The authors declare that there are no conflicts of interest regarding the publication of this paper.	2019-08-02T20:21:52.732Z	2019-07-18T00:00:00.000	{ "year": 2019, "sha1": "44cbbb911658dcd51181755f11d2e4bf5686f8ac", "oa_license": "CCBY", "oa_url": "https://downloads.hindawi.com/journals/mpe/2019/9535294.pdf", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "d3e9c1641ba843e38e70140fc27bdc7ea235f584", "s2fieldsofstudy": [ "Physics", "Mathematics" ], "extfieldsofstudy": [ "Physics" ] }
231593217	pes2o/s2orc	v3-fos-license	MicroRNA-21 Contributes to Cutaneous Squamous Cell Carcinoma Progression via Mediating TIMP3/PI3K/AKT Signaling Axis Background Though the therapeutic potentials of microRNAs (miRNAs) are extensively explored in cutaneous squamous cell carcinoma (CSCC), the concrete function of miR-21 in this disorder has not been thoroughly comprehended. Therein, this work is launched to clarify the miR-21-pivoted mechanism in CSCC from the perspective of tissue inhibitor of metalloproteinases-3 (TIMP3) and phosphatidylinositol 3-kinase (PI3K)/protein kinase B (AKT) pathway. Methods Microarray-based analysis was utilized to screen out miR-21 with the most up-regulated expression in CSCC tissues. The relation between miR-21 and TIMP3 expression in tissues, and the overall survival of CSCC patients was evaluated. Loss-of-function assays were performed in cells to explore the independent and combined functions of miR-21 and TIMP3 in CSCC cell progression. Mice were injected with miR-21 inhibitor or TIMP3 si for identifying their roles in tumor formation and liver metastasis. The mechanism among miR-21, TIMP3 and PI3K/AKT pathway was interpreted. Results MiR-21 was up-regulated while TIMP3 was down-regulated in CSCC tissues, which were connected with unsatisfactory survival of patients. Down-regulating miR-21 inhibited CSCC cell progression and retarded CSCC tumor formation and metastasis in mice. Silencing of TIMP3 reversed the effects of miR-21 down-regulation on CSCC progression. Besides, down-regulating miR-21 inhibited PI3K/AKT pathway activation in CSCC cells via mediating TIMP3. Conclusion It is elucidated that miR-21 depletion impedes CSCC cell invasion and metastasis via enhancing TIMP3 and suppressing PI3K/AKT pathway activation. Introduction Cutaneous squamous cell carcinoma (CSCC) is the second most overwhelming non-melanoma skin cancer. 1 The incidence of CSCC is rising with high potential morbidity and mortality, especially in the elderly and immunocompromised people. 2 As to CSCC management in different situations, surgery is recommended for elderly patients with well-differentiated tumors, adjuvant radiation therapy for incomplete resection, radical dissection for patients with positive lymph nodes, while PD-1 inhibitor for advanced CSCC. 3 However, during the period of treatment, recurrence commonly happens in CSCC. 4 Concerning the severity and complexity of this disease, the need for effective agents is still a major task. As a component of microRNA (miRNA), miR-21 has been discussed to own diagnostic, therapeutic and prognostic effects on cutaneous melanoma and CSCC. 5 miRNAs are involved in the initiation, migration, invasion, and chemoresistance in CSCC, and modulate the cellular responses by acting as anti-tumor factors. 6 The significance of miR-21 overexpression in CSCC tumorigenesis has been evaluated in multiple studies. Clinically and experimentally, the resistant CSCC is characterized by elevated miR-21 in patients and mice. 7 Mechanistically, miR-21 has been revealed to be induced in invasive CSCC, harboring diagnostic values. 8 Furthermore, miR-21 has been identified as a pro-oncogene in CSCC and it is overexpressed in CSCC. 9 The binding relationship between miR-21 and tissue inhibitor of metalloproteinases-3 (TIMP3) has been highlighted in cervical cancer. 10 TIMP3, an extracellular matrix-bound protein has been documented to participate in skin cancers. 11 As documented in a study, it is presented that TIMP3 is downregulated in melanoma progression and it is a negative regulator for cell migration and invasion, as well as anoikis resistance. 12 In addition, there is a study specifically indicating that TIMP3 expression is closely related to angiogenesis in malignant melanoma. 11 miR-21 is clarified to mediate TIMP3, and miR-21 overexpression results in a reduction in TIMP3 expression, thereafter enhancing melanoma cell invasion. 13 However, how their interaction functions in CSCC remain undetermined. Intriguingly, the dual-luciferase reporter gene assay in the present study identified the direct binding relationship between miR-21 and TIMP3. Furthermore, phosphatidylinositol 3-kinase (PI3K)/protein kinase B (AKT) pathway has been evidenced to be involved in the pathogenesis of CSCC. 14 Notably, another study has proved that the PI3K/AKT serves as a mediator in CSCC cell growth and development. 15 On top of that, the combined interplay of miR-21, TIMP3 and PI3K/AKT pathway in CSCC is lack of explorations. Thereby, this work is projected to decode the functions of miR-21/TIMP3/PI3K/AKT pathway axis in CSCC. Clinical Data Tumor tissues and adjacent tissues were collected from 56 CSCC patients (35 male and 21 female) admitted to the Third Affiliated Hospital of Qiqihar Medical University from June 2013 to June 2016. This study was explicitly approved by the Ethics Committee of the Third Affiliated Hospital of Qiqihar Medical University which followed principles in the Declaration of Helsinki. Written informed consents were obtained from all participants. The tissue specimens were preserved in liquid nitrogen at −80°C. SCC Borders Pathological Classification had clarified CSCC into stage I (well-differentiated) with 0-5% undifferentiated cells, stages II-III (moderately differentiated) with 25-75% undifferentiated cells and stage IV (poorly differentiated) with 75-100% undifferentiated cells. In line with the criteria, 39 cases were well-differentiated, 15 cases were moderately differentiated, and 2 cases were poorly differentiated. The detail information of patients is listed in Table 1. Microarray-Based Analysis The tumor and adjacent tissue specimens from CSCC patients were ground with liquid nitrogen and supplemented with Trizol reagent (Thermo Fisher Scientific, Waltham, MA, USA) for total RAN extraction. The RNA purity and integrity were detected by an ultraviolet spectrophotometer and formaldehyde denaturation agarose gel. A miRCURYTM Array Power Labeling Kit (Exiqon, Colony Formation Assay Cells were transfected with low expression vectors and incubated to stable growth. Detached by 0.25% trypsin, cells (5 × 10 3 ) were seeded in a 6-well plate containing 1 mL medium, with the medium renewed every 3 days. After a 10-day culture, colonies were visible, and the cells were fixed with 5 mL methanol and stained with 0.5% crystal violet solution. The number of colonies (>10 cells) was calculated, and the rate of colony formation = number of colonies/number of seeded cells × 100%. Scratch Test Seeded in a 6-well plate, cells (1 × 10 6 ) were supplemented with medium and cultured to 90% confluence. Scratches were made by a 200-mL pipette, and cell debris were washed with phosphate-buffered saline. Incubated for 5 hours, cells were observed under an inverted microscope (Eclipse Ti, Nikon, Tokyo, Japan) in 5 fields of view. Cell migration ability was detected by wound healing distance. Transwell Assay Suspended in 100 µL DMEM to reach 1 × 10 5 cells/mL, cells were seeded into the apical Transwell chamber with 8-µM pore size and added with Matrigel (Corning, NY, USA). DMEM (200 µL) containing 15% FBS was added to the basolateral Transwell chamber. Following a 24-hour incubation, the cells in the apical chamber were wiped, and those in the basolateral chamber were fixed in 4% paraformaldehyde, stained with 0.5% crystal violet solution and observed under an inverted microscope (Eclipse Ti, Nikon) in 5 five fields of view. Cell invasion was Hematoxylin-Eosin (HE) Staining The mouse liver tissues were fixed with 4% paraformaldehyde overnight, embedded, sectioned and stained. Dewaxed in xylene, the sections were soaked in gradient alcohol, stained with hematoxylin staining solution (Solarbio) for 5 min and differentiated with 1% hydrochloric acid for 3 s. Following that, the sections were stained by 5% eosin stain (Solarbio) for 3 min, dehydrated and sealed in neutral gum. The sections were observed and photographed under an inverted microscope (Eclipse Ti, Nikon) in five fields of view. The number and area of liver nodules were calculated to detect cell metastatic ability in vivo. MiR-21 is Up-Regulated in CSCC Tissues and Cells In order to find the miRNA highly expressed in CSCC, miRNA microarray analysis was performed on tumor tissues and adjacent tissues of three randomly selected CSCC patients. It was manifested that eight miRNAs were upregulated in tumor tissues, among which miR-21 was the most up-regulated one ( Figure 1A). Further detection of miR-21 expression in all CSCC patients was carried out, and the results exhibited that miR-21 was up-regulated in CSCC patients ( Figure 1B). Survival analysis revealed that patients with higher miR-21 had a lower survival rate and shorter survival time than those with lower miR-21 ( Figure 1C). miR-21 expression in HSC-1, A431 and HaCaT cells was measured, with the results indicating that miR-21 was elevated in HSC-1 and A431 cells versus HaCaT cells ( Figure 1D). Down-Regulating miR-21 Reverts the Malignant Phenotype of CSCC Cells Afterwards, miR-21 down-regulation was introduced in HSC-1 and A431 cells. Three miR-21 inhibitor fragments were synthesized for transfection. RT-qPCR implied that miR-21 inhibitor#1 demonstrated the highest efficiency, which was selected for the subsequent analysis ( Figure 2A). After miR-21 knockdown in HSC-1 and A431 cells, cells were cultured for the same period under the same conditions. Cultured to the 10th day, the colony formation of cells with poorly expressed miR-21 was diminished ( Figure 2B). EdU assay demonstrated that EdU-positive rate reduced in cells with poor miR-21 expression ( Figure 2C). Scratch test observed that after 24 hours of incubation, the wound healing rate of cells with depleted miR-21 was slower ( Figure 2D). Seeded in the Transwell chamber and cultured for 24 hours, it was depicted that cells with down-regulated Figure 2E). Flow cytometry of cell cycle distribution and apoptosis implied that miR-21 down-regulation arrested most cells in the G0/G1 phase and reduced cells in the S phase ( Figure 2F), but also increased apoptotic cells and impaired cell viability ( Figure 2G). It could be summarized that miR-21 down-regulation partially prevented CSCC cells from proliferating, invading and migrating and advanced apoptosis. Down-Regulating miR-21 Retards CSCC Tumor Formation and Liver Metastasis in Mice Mice were injected with cells transfected with miR-21 inhibitor or its NC, and tumor volume was measured every 7 days. It was outlined that the tumor grew slowly in mice injected with miR-21 inhibitor-transfected cell ( Figure 3A). About 28 days later, the tumors were harvested from mice and weighed, and depleted miR-21 reduced tumor weight ( Figure 3B). After intravenous injection of cells with miR-21 inhibitor, the livers of mice on the 30th day were stained and observed, with the findings picturing that miR-21 depletion reduced the number of liver nodules ( Figure 3C). The expression of miR-21 in mouse xenograft tumors was examined and found to be significantly reduced in the tumors harboring miR-21 inhibitor ( Figure 3D), indicating that downregulation of miR-21 inhibits CSCC growth and metastasis in vivo. TIMP3 is Targeted by miR-21 It was predicted by StarBase, TargetScan, miRDB, miRwalk, miRBase and miRanda that TIMP3 was targeted by miR-21 ( Figure 4A). Dual-luciferase report gene assay was adopted to validate the targeting relationship between TIMP3 and miR-21, and the results elucidated that the fluorescence intensity of cells co-transfected with miR-21 inhibitor and TIMP3-WT was enhanced ( Figure 4B). Detection of TIMP3 expression in tumor tissues and adjacent tissues revealed that TIMP3 was down-regulated in CSCC patients ( Figure 4C). Correlation analysis of miR-21 and TIMP3 expression in CSCC tissues of patients suggested a negative connection ( Figure 4D). In terms of the linkage between TIMP3 expression and overall survival of CSCC patients, the results illustrated that TIMP3 expression at a high level contributed to a higher survival rate and longer survival time than patients with TIMP3 at a low expression ( Figure 4E). TIMP3 down-regulation was administrated into cells, and TIMP3 expression was the lowest in the second fragment ( Figure 4F). TIMP3 down-regulation resulted in enhanced cell invasion and migration abilities in Transwell assays ( Figure 4G). Furthermore, the poor expression of TIMP3 increased liver nodules number and enlarged liver nodules area ( Figure 4H). Silencing of TIMP3 Reverses the Effects of miR-21 Down-Regulation on CSCC CSCC cells were transfected with miR-21 inhibitor alone or with TIMP3 si. To verify the transfection efficiency of TIMP3 si and its correlation with miR-21, we examined the protein expression of TIMP3 in cells after transfection. The results of Western blot revealed that miR-21 inhibitor significantly increased TIMP3 protein expression in cells, while co-transfection of TIMP3 si effectively suppressed its expression ( Figure 5A). TIMP3 down-regulation antagonized the inhibitory effects of miR-21 down-regulation on colonyforming ability, migration distance and invasion and the promoting effect on cell apoptosis ( Figure 5B-E). TIMP3 down-regulation followed by miR-21 depletion increased liver nodules and promoted metastasis in mice ( Figure 5F). Down-Regulating miR-21 Impairs the PI3K/AKT Signaling Pathway Activation in CSCC The PI3K/AKT pathway has been evidenced to be involved in the pathogenesis of CSCC. 12 Notably, another study has proved that the PI3K/AKT pathway serves as a mediator in CSCC cell growth and development. 13 Therefore, the phosphorylation levels of PI3K and AKT in CSCC cells were examined, and the results suggested that PI3K/AKT pathway was activated ( Figure 6A). Due to the higher activation of PI3K/AKT pathway in HSC-1 cells, HSC-1 cells transfected with miR-21 inhibitor were utilized for subsequent Western blot for PI3K/AKT pathway activity assessment. It was demonstrated that miR-21 DovePress suppression hindered the activation of PI3K/AKT pathway. It indicated that miR-21 affected PI3K/AKT pathway activation, while changes in the PI3K/AKT pathway activity are also responsible for the changes in CSCC cell activity caused by miR-21 downregulation ( Figure 6B). In cells transfected with TIMP3 si, the phosphorylation levels of PI3K and AKT were increased, suggesting that the effect of TIMP3 on cell activity is caused by impairment of the PI3K/AKT pathway ( Figure 6C). Also, PI3K/ AKT pathway activation was reinforced after cotransfected with miR-21 inhibitor and TIMP3 si ( Figure 6D), which further proved the targeting relationship between miR-21 and TIMP3. The above results indicated that miR-21 mediated PI3K/AKT pathway by regulating TIMP3 in CSCC. Discussion Ranked as the most prevalent cancers among the white populations, CSCC imposes threatens on human health with metastatic potentials. 16 Furthermore, miRNAs are participated in modulating the expression of cancerrelated genes by enhancing the initiation, development, invasiveness, and aggressiveness of CSCC, making them potential prognostic biomarkers and therapeutic candidates in CSCC target therapy. 17 Enlightened by the previous studies, which have validated the potent roles of miR-21 in CSCC, this work is launched with the results concluding that miR-21 overexpression facilitated the development of CSCC through down-regulating TIMP3 and potentiating PI3K/AKT pathway activation. Initially, determinations of miR-21 in clinical CSCC tissues were implemented with the results stratifying that miR-21 expression was elevated in CSCC tissues and cells, and miR-21 expression was negatively associated with the overall survival of CSCC patients. Then, miR-21 down-regulation assays were performed on CSCC cells, and it was manifested that miR-21 depletion reverted the malignant phenotype of CSCC cells. For further validation, cell transfected with miR-21 inhibitor was injected into mice, and the treated mice were featured by suppressed tumor growth and liver metastasis. As depicted in former studies, it is supportive that miR-21 expression is up-regulated in CSCC. 8,9,18 Actually, miR-21 upregulation is documented to promote cell migration in vitro and metastasis in vivo in melanoma. 19 Moreover, it is surveyed that miR-21 reaches a high level in primary melanoma tissues and miR-21 elimination induces melanoma cell apoptosis. 20 Intriguingly, overexpressed miR-21 is connected with advanced clinical stage and unsatisfactory 5-year overall survival, and depleting miR-21 is beneficial for apoptosis and chemo-or radiosensitivity in human cutaneous malignant melanoma. 21 Echoed with the findings in this work, miR-21 expression is raised in breast cancer tissues, which is connected with dismal survival rate. 22 All these data suggest that miR-21 plays an oncogenic role in human diseases. Followed by that, online websites and dual-luciferase reporter gene assays have predicted and verified that miR-21 targeted and negatively connected with TIMP3. Also, TIMP3 was down-regulated in clinical CSCC tissues, which also was linked to the unsatisfactory survival of patients. TIMP3 down-regulation assays were conducted in cells and mice, and the results demonstrated that TIMP3 knockdown promoted CSCC cell invasion and liver metastasis in mice. Exactly, there are studies identifying the targeting connection between miR-21 and TIMP3 23,24 which are supportive to our finding. Mechanistically, TIMP3 elevation by down-regulating miR-21 accredits to decreased colon cancer cell invasion and metastasis in vitro and in vivo. 25 A reduction can be seen in TIMP3 expression in melanoma which is related to overall and disease-free survival, and its restoration sets obstructions on the way of melanoma cell invasion and migration. 12 Besides, a conclusion drawn from a previous study has suggested that TIMP3 inhibition induced by elevated miR-21 results in increments in the invasiveness of melanoma cells. 13 Briefly, the listed studies have confirmed the results concluded in this work. Subsequently, for a thorough comprehension of the involvement of miR-21/TIMP3 axis in CSCC, the extent of PI3K and AKT phosphorylation were detected. The findings elucidated that miR-21 inhibition disrupted the PI3K/AKT pathway activation, as evidenced by the reduced extent of PI3K and AKT phosphorylation, which was antagonized by down-regulating TIMP3, indicating that miR-21 targeted TIMP3 to mediate the PI3K/AKT pathway activation. The activated PI3K/AKT pathway has been mentioned in skin cancer carcinogenesis, 26 and the PI3K/AKT pathway deficit can partially attenuate melanoma progression. 27 Moreover, the PI3K/AKT/mTOR pathway was displayed by Nardo et al to play an indispensable role in the pathogenesis of CSCC. 14 A recent study indicated that miR-21 directly targets and inhibits the expression of PTEN (a negative modulator of the PI3K/AKT pathway), and miR-21 inhibition upregulated PTEN expression but impaired the PI3K/AKT pathway, thereby elevating liver cancer cell apoptosis. 28 Furthermore, miR-21 down-regulation causes an DovePress impairment in PI3K/AKT pathway activation in Burkitt's lymphoma. 29 Also, PI3K/AKT pathway activation is in part suppressed by up-regulating TIMP3 in oral squamous cell carcinoma. 30 Anyway, the aforementioned research presentations are echoed with the discoveries in this work. Conclusion In summary, this work has elaborated that restoring of miR-21 or silencing of TIMP3 deteriorates CSCC by activating the PI3K/AKT pathway (Figure 7), which replenishes the exited knowledge about CSCC-oriented mechanism. However, limited by the relatively small experimental scale, a large cohort of researches are in need of further confirmation of the results concluded in this study. Funding There is no funding to report. Disclosure The authors declare that they have no competing interests.	2021-01-14T05:05:12.237Z	2021-01-01T00:00:00.000	{ "year": 2021, "sha1": "77e4a13147d0f4a7ac5d6bc2b5e134157c3d00f9", "oa_license": "CCBYNC", "oa_url": "https://www.dovepress.com/getfile.php?fileID=65517", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "77e4a13147d0f4a7ac5d6bc2b5e134157c3d00f9", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
3800610	pes2o/s2orc	v3-fos-license	Arc/Arg3.1 protein expression in dorsal hippocampal CA1, a candidate event as a biomarker for the effects of exercise on chronic stress-evoked behavioral abnormalities [Purpose] Chronic stress is a risk factor for behavioral deficits, including impaired memory processing and depression. Exercise is well known to have beneficial impacts on brain health. [Methods] Mice were forced to treadmill running (4-week) during chronic restraint stress (6h/21d), and then behavioral tests were conducted by Novel object recognition, forced swimming test: FST, sociality test: SI. Dissected brain was stained with anti-calbindin-d28k and anti-Arc antibodies. Also, mice were treated with CX546 intraperitoneally during chronic restraint stress, and behavioral tests were assessed using Morris water maze, FST, and SI. Dissected brain was stained with anti-Arc antibody. [Results] The current study demonstrated that chronic stress-induced impairment of memory consolidation and depression-like behaviors, along with the changes in calbindin-d28k and Arc protein levels in the hippocampal CA1 area, were attenuated by regular treadmill running. Further, prolonged ampakine treatment prevented chronic stress-evoked behavioral abnormalities and nuclear Arc levels in hippocampal CA1 neurons. Nuclear localization of Arc protein in hippocampal CA1 neurons, but not total levels, was correlated with behavioral outcome in chronically stressed mice in response to a regular exercise regimen. [Conclusion] These results suggest that nuclear levels of Arc are strongly associated with behavioral changes, and highlight the role of exercise acting through an α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor (AMPAR)-mediated mechanisms in a chronic stress-induced maladaptive condition. INTRODUCTION Chronic stress is a potent predictor for the development of memory processing impairments and psychiatric defects, including longterm memory impairments and depression. In particular, chronic stress-evoked abnormalities in activity-dependent synaptic plasticity are strongly associated with these neuropathophysiological conditions 1 . The activity-regulated cytoskeletal gene Arc, also known as Arg3.1, plays a critical role in synaptic strength and plasticity during memory consolidation. Its transcription, translation, localization, and stability are tightly controlled by neuronal activity 2 , suggesting various functions depending on its particular characteristics. In fact, Arc plays a pivotal role in long-term potentiation (LTP) by regulating cytoskeletal dynamics and spine morphology, as well as in long-term depression (LTD) by mediating metabotropic glutamate receptor (mGluR)-induced LTD via α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor (AMPAR) endocytosis at synapses [3][4] . Several studies have demonstrated that Arc expression mostly increases in response to acute stress in the prefrontal cortex, with an AMPAR and N-methyl-D-aspartate (NMDA) receptor (NMDAR)-dependent increase in glutamatergic neurotransmission 5-6. In particular, cortical AMPAR expression and AMPAR-mediated excitatory postsynaptic currents (EPSCs) are induced by various acute stressors 6-7 . Specific chronic stress-induced changes of hippocampal Arc expression remain unclear; conflicting studies have shown that hippocampal Arc expression is either increased or decreased by chronic stress [8][9] . Arc protein localizes to active synapses and nuclei, though its function has been extensively explored at the synapse. Translated Arc protein translocates into the nucleus, and contributes to GluR1 expression and synaptic plasticity in stimulated neurons, in in vivo and in vitro experiments 10-11 . Various forms of exercise have the ability to prevent, restore, or ameliorate chronic stress-induced brain disorders, including Arc/Arg3.1 protein expression in dorsal hippocampal CA1, a candidate event as a biomarker for the effects of exercise on chronic stress-evoked behavioral abnormalities Yea-Hyun Leem 1 / Hyukki Chang 1 * [Purpose] Chronic stress is a risk factor for behavioral deficits, including impaired memory processing and depression. Exercise is well known to have beneficial impacts on brain health. [Methods] Mice were forced to treadmill running (4-week) during chronic restraint stress (6h/21d), and then behavioral tests were conducted by Novel object recognition, forced swimming test: FST, sociality test: SI. Dissected brain was stained with anti-calbindin-d28k and anti-Arc antibodies. Also, mice were treated with CX546 intraperitoneally during chronic restraint stress, and behavioral tests were assessed using Morris water maze, FST, and SI. Dissected brain was stained with anti-Arc antibody. [Results] The current study demonstrated that chronic stress-induced impairment of memory consolidation and depression-like behaviors, along with the changes in calbindin-d28k and Arc protein levels in the hippocampal CA1 area, were attenuated by regular treadmill running. Further, prolonged ampakine treatment prevented chronic stress-evoked behavioral abnormalities and nuclear Arc levels in hippocampal CA1 neurons. Nuclear localization of Arc protein in hippocampal CA1 neurons, but not total levels, was correlated with behavioral outcome in chronically stressed mice in response to a regular exercise regimen. [Conclusion] These results suggest that nuclear levels of Arc are strongly associated with behavioral changes, and highlight the role of exercise acting through an α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor (AMPAR)-mediated mechanisms in a chronic stress-induced maladaptive condition. [Key words] Arc, chronic stress, memory consolidation, depression, regular exercise, AMPA receptor cognitive impairment and depression 12-13 . For example, chronic stress-induced impairments in learning and memory are ameliorated by regular exercise via cAMP-activated protein kinase (AMPK)-dependent BDNF activity in chronically stressed mice 12 . Chronic stress-induced abnormalities in glutamate transmission and synaptic plasticity are associated with AMPAR function, as evidenced by a decrease in excitation of temporoammonic (TA)-CA1 path synapses, and a decrease in AMPAR expression in the hippocampal CA1 region 13-14 . On the contrary, 4 weeks of voluntary wheel running, (but not acute exercise) enhances GluR1 and pGluR1 (Ser845) levels in the hippocampus 15 . Repeated exercise alters AMPAR subunit distribution in diverse brain regions, evidenced by changes in AMPAR subunits in the sensory-motor cortex, cerebellum, and striatum 16 . However, the nuclear localization of Arc protein in hippocampal CA1 neurons, the behavioral outcomes of chronic stress, and the effects of regular exercise on such factors are under-explored. The present study demonstrates a correlation between chronic stress-induced alterations in nuclear Arc levels in hippocampal CA1 neurons and AMPAR activity, and explores the potential role of exercise. Experimental mice Male 7-week-old C57BL/6 mice were obtained from Dae Han Biolink Co., Ltd. (Eumsung, Chungbuk, Korea), and all experimental procedures involving animals were approved by the Animal Care and Use Committee at Seoul Women's University. Behavioral tests Depressive-like behavior (N=7-8/group) was assessed using two methods: the forced swimming test (FST) and the sociality test 17 . For the FST, all mice were exposed to a 15-min pre-test on day 1. The test was then conducted 24 h later and recorded using a video camera. Mice were forced to swim for 6 min, and the time spent immobile was measured after the first min. Briefly, for the sociality test, an apparatus was partitioned into three equal chambers with clear Plexiglas dividing walls that could be removed to allow free access to each chamber. Before testing, each mouse was first acclimatized by being placed into the closed-off center compartment for 5 min. An unfamiliar conspecific male mouse that had no prior contact with the test mouse was enclosed in a wire cup in either the third spot from the left or the right chamber for 1 min, following which the test mouse was allowed to explore for 5 min by removing both dividing walls. Sociality was quantified as the time spent in the interaction zone (IZ) with the cup containing the novel male, or the non-interaction zone (Non-IZ) containing an empty cup. Social interaction (SI) index was calculated using the following equation: SI (%) = IZ/(non-IZ + IZ) x 100. Memory consolidation was evaluated using two methods: the Morris water maze (MWM) 18 and the novel object recognition test 19 . For the MWM, mice were randomly assigned a quadrant (NW, NE, SE, or SW) located in a 1.5-m-diameter black circular pool containing 22°C water that contained a stationary escape platform. Probe tests were conducted 24 h after the last 10-trial training block as well as 28 days later, wherein the platform was removed and mice swam from a random start location for 60 s. The time spent in the target quadrant (that used to contain the platform) and the latency to the target were determined using SMART 3.0 software (Panlab, S.L.U., Barcelona, Spain) on a computer connected to a ceiling-mounted camera directly above the pool. For the novel object recognition test, mice were handled and habituated to an empty open field arena for 2 days. An acquisition session was performed, wherein mice were placed into the arena with two identical objects (Falcon tissue culture flasks filled with sand) that were positioned in two corners of the apparatus, and allowed to explore the objects for 10 mins. Following a 3 day delay, one of the familiar objects was replaced with a novel object (interlocking tower of Lego pieces with different shapes and colors), and mice were allowed to explore for 5 mins. The time spent exploring each object was recorded using a video-tracking system during the acquisition and retention phases. The recognition ratio (%) was expressed as the percentage of time spent with each object compared with the total exploration time. To facilitate sustained AMPAR activation, CX546 [0-20 mg/kg CX546 dissolved in 16.5% 2-hydroxypropyl-bcyclododextrin (CDX) in 0.9% saline, Abcam, Cambridge, UK], a potent ampakine, was injected intraperitoneally twice daily every 2 days during the period of stress exposure, a dose regimen chosen based on preliminary work wherein this paradigm increased BDNF protein levels without altering exploratory behavior 20 . Immunohistochemical analyses Anesthetized mice were perfused with 100 mM PBS (pH 7.4), followed by cold 4% paraformaldehyde in PBS. Every fourth section (40 µm-thickness) was taken from the region between bregma -1.82 mm and -2.18 mm. Free-floating staining was conducted using traditional methods. Sections were incubated overnight at 4°C with anti-calbindin-D-28k (CalB) and anti-Arc/Arg3.1 antibody (Abcam, Cambridge, rabbit polyclonal, 1: 2,000) then with biotinylated secondary antibodies (Vector Laboratories; Burlingame, CA, USA 1: 200, respectively), followed by visualization using the ABC and DAB methods (ABC Elite kit, Vector Laboratories; Burlingame, CA, USA). Sections were slide mounted and digitally imaged (captured at 100× magnification), then analyzed using Image J software (NIH Image Engineering, Bethesda, MD). The number of positive cells per mm2 was counted within a defined circle (each group N=4-5). Data are presented as means ± SD (n = 10-12 animals). * and ** denote significant differences at p < 0.05 and p < 0.01, respectively. Statistical analyses Significant differences between groups were determined using paired t-tests and one-way analyses of variance (ANOVA; SPSS for Windows, version 18.0, Chicago, IL, USA). Post-hoc comparisons were performed using Student-Newman-Keuls tests. All values are reported as mean ± standard deviation (SD). Values of p < 0.05 were considered statistically significant. RESULTS Regular exercise prevented chronic stress-induced failure of memory consolidation and behavioral depression, with a simultaneous change in calbindin levels as well as total and nuclear Arc protein levels in hippocampal CA1 neurons. We found that chronic stress (6h/21d) induced a decrease in the exploration ratio of the novel object, a decrease in SI index, and an increase in immobility during the FST, which was reversed by regular exercise (Fig. 1A There were fewer CalB+ hippocampal CA1 cells as a result of chronic stress, and this decrease was restored to control levels by exercise ( Fig. 1C; F3, 12 = 12.12, p < 0.01). There was a profound enhancement of Arc+ hippocampal CA1 cells following chronic stress, regardless of treadmill running or exercise alone regimens ( Fig. 1D; F3, 12 = 11.99, p < 0.01). Chronic stress-induced enhancement of the ratio of nuclear to cytoplasmic levels of Arc was restored to basal levels by exercise, and the nuclear ratio of Arc in exercise alone mice was comparable to that of controls (Fig. 1D right panel; F3, 12 = 8.54, p < 0.01). Prolonged ampakine treatment prevented chronic stress-induced failure of memory consolidation and behavioral depression, with a simultaneous change in total and nuclear Arc protein levels in hippocampal CA1 neurons. A 10-block training paradigm was used in the MWM, which resulted in the successful consolidation of longterm memory ( Fig. 2A In the sociality test, chronic stress reduced the SI index, and this decrease was attenuated by CX546 administration (20 mg/kg; Fig. 2Ca; F4, 30 = 2.63, p < 0.05). Immobility in the FST was enhanced by chronic stress, and this increase was reversed by CX546 (10-20 mg/kg; Fig. 2Cb; F4, 30 = 4.98, p < 0.01). The immunoreactivity of Arc in hippocampal CA1 cells was profoundly enhanced by chronic stress, regardless of CX546 treatment ( Fig. 1D; F4, 20 = 10.31, p < 0.01). Chronic stress-induced enhancement of the nuclear localization ratio of Arc (i.e., the ratio of nuclear to cytoplasmic levels) returned to basal level following CX546 treatment (20 mg/kg; Fig. 1D DISCUSSION The current study demonstrated that regular exercise exerted protective effects against chronic stress-induced behavioral deficits, including the impairment of longterm memory formation and the development of depressive symptoms, likely by reducing AMPAR-mediated excitatory responsiveness in hippocampal CA1 cells. The 21 consecutive day restraint stress model resulted in impaired memory retention as detected by the MWM and novel object recognition tests, as well as behavioral depression measured by SI index and FST, which were restored by 4 weeks of regular treadmill running. These results suggest that this experimental paradigm is valid for exploring the mechanisms underlying improvement or prevention of chronic stress-induced behavioral deficits in memory processing and mood-related illnesses using regular exercise. Calbindin d28k, a calcium binding protein, contributes to intracellular Ca2+ homeostasis by buffering excess intracellular calcium and maintaining intracellular Ca2+ levels to prevent depletion 22-23 . Calbindin d28k protein expression in hippocampal CA1 cells was markedly reduced by chronic stress, and this decrease was attenuated by exercise (Fig. 1C). This suggests that chronic stress is related to disrupted intracellular calcium homeostasis, and that this process is affected by exercise. Several studies have suggested that calcium channel-dependent depolarization and kinases such as CaMKs play a crucial role in synaptic plasticity in chronic stress-induced behavioral deficits, and that the effects of exercise work through Ca2+/Calmodulin-dependent Kinase II-dependent BDNF and CREB induction to alter synaptic plasticity [24][25][26] . Previous studies support these results and have shown that disrupted calcium homeostasis and signaling by chronic stress may be overcome by regular exercise. Lower field excitatory postsynaptic potential (fEPSP) slopes and reduced GluA1 expression in hippocampal CA1 neurons have been observed following chronic stress 13 . Although synaptic Ca2+ entry occurs predominantly through NMDARs, activity-dependent synaptic Ca2+ currents by AMPARs are also critical for synaptic plasticity, for example during hippocampal NMDAR-dependent LTP 13-14 . Arc induction takes place as part of Data are presented as means ± SD (n = 10-12 animals). * and ** denote differences at p < 0.05 and p < 0.01, respectively. The role of Arc in chronic stress and regular exercise Journal of Exercise Nutrition & Biochemistry the second wave of molecules that are modulated by c-fos for late-LTP and memory consolidation, and contributes to AMPAR trafficking, synaptic strengthening, and long-term neuronal plasticity at excitatory synapses 4, 27 . Our results, along with previous findings, led us to assess Arc expression in hippocampal CA1. Hippocampal CA1 Arc expression was profoundly enhanced by chronic stress, regardless of exercise regimen, while nuclear localization of Arc was significantly increased by chronic stress, and reverted to the basal state following exercise (Fig. 1D). Recent evidence from in vivo and in vitro studies has demonstrated that a rapid increase in cytoplasmic Arc protein slowly translocates into the nucleus where it associates with nuclear promyelocytic leukemia (PML) and contributes to AMPAR trafficking and homeostatic plasticity in stimulated neurons 10-11 . To elucidate the direct relationship between AMPA receptor activity and nuclear localization of Arc protein under chronic stressful condition, CX546, an ampakine that produces delayed desensitization of AMPA receptors, AMPAR-mediated increases in amplitude, and duration of fast, excitatory transmission 28 , was injected into mice subjected to chronic stress. Nuclear Arc levels in hippocampal CA1 neurons were augmented by stress. This effect was attenuated with CX546 treatment, with corresponding normalization of behavior. Nuclear PML bodies contribute to transcription and mRNA export, and nuclear Arc expression reduces GluR1 transcription and abnormal homeostatic response to increased activity 1, 12 . Based on previous studies, the increase in nuclear Arc protein attenuates the surface trafficking of AMPA receptors by reducing GluR1 levels, which produces AMPAR-mediated downregulation of dendritic-wide homeostatic scaling under maladaptive conditions such as chronic stress. On the contrary, regular exercise is believed to prevent chronic stress-related neurophysiological and behavioral changes. In addition, we suggest that the quantification of nuclear translocation of Arc in hippocampal CA1 cells could act as a biomarker for stress-related behavioral deficits.	2018-04-03T03:42:07.390Z	2017-12-31T00:00:00.000	{ "year": 2017, "sha1": "06e87374584188c383e9f0addd8298744f44a9b0", "oa_license": "CCBY", "oa_url": "https://doi.org/10.20463/jenb.2017.0033", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "084bb9f21f56cb51a0ea38d4197c2b5cabc04f44", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
257057731	pes2o/s2orc	v3-fos-license	A preclinical radiotherapy dosimetry audit using a realistic 3D printed murine phantom Preclinical radiation research lacks standardized dosimetry procedures that provide traceability to a primary standard. Consequently, ensuring accuracy and reproducibility between studies is challenging. Using 3D printed murine phantoms we undertook a dosimetry audit of Xstrahl Small Animal Radiation Research Platforms (SARRPs) installed at 7 UK centres. The geometrically realistic phantom accommodated alanine pellets and Gafchromic EBT3 film for simultaneous measurement of the dose delivered and the dose distribution within a 2D plane, respectively. Two irradiation scenarios were developed: (1) a 10 × 10 mm2 static field targeting the pelvis, and (2) a 5 × 5 mm2 90° arc targeting the brain. For static fields, the absolute difference between the planned dose and alanine measurement across all centres was 4.1 ± 4.3% (mean ± standard deviation), with an overall range of − 2.3 to 10.5%. For arc fields, the difference was − 1.2% ± 6.1%, with a range of − 13.1 to 7.7%. EBT3 dose measurements were greater than alanine by 2.0 ± 2.5% and 3.5 ± 6.0% (mean ± standard deviation) for the static and arc fields, respectively. 2D dose distributions showed discrepancies to the planned dose at the field edges. The audit demonstrates that further work on preclinical radiotherapy quality assurance processes is merited. in the comparability and reproducibility of the results. In the present study, we used a postal audit to investigate dosimetric conformance of UK image-guided small animal precision irradiation facilities. Measurements of the dose delivered for both static and arc beam geometries were made using a realistic murine phantom containing Gafchromic EBT3 film (Vertec Scientific Ltd. Reading, UK) and alanine dosimeters (NPL, Middlesex, UK), and were compared against the dose predicted by the treatment planning systems (TPS). We additionally surveyed participating institutes to ascertain the equipment and techniques currently supported by departments, quality control processes in use, and attitudes to the need for quality control in pre-clinical radiation experiments. Methods Questionnaire. Prior to the audit, information regarding the equipment in use, the techniques implemented and the QA procedures in place at 7 centres across the UK which actively undertake in vivo radiation research using an Xstrahl (Walsall, UK) Small Animal Radiation Research Platform (SARRP) was gathered via a questionnaire. The questionnaire also invited participants to give their opinion, through open-ended questions, on the necessity of dosimetry audits and defined protocols, the level of acceptable dose tolerances, QA responsibilities and required reporting parameters in publications to improve the currently poor reproducibility of research 18 . Phantom design and printing. The phantom used in this audit was previously described by Price et al. 22 (which includes a link to open source files for the phantom geometry). To create the phantom the cone beam computed tomography (CBCT) scan of a nude mouse was segmented into three parts (body, bones and lungs) then transformed into stereolithographic files suitable for import into Meshmixer (Autodesk, Inc.) and Netfabb (Autodesk, Inc.) computer-aided design software. In this study, the phantom (body and lungs) was split on the central coronal plane to accommodate Gafchromic EBT3 film, and a cylindrical cavity 6 mm in diameter and 2.5 mm in height was incorporated in the ventral half of the phantom, in the brain region, to contain an alanine detector. The use of a higher density bone-equivalent material was excluded from this study to avoid any dosimetric uncertainties relating to the tissue segmentation process within the SARRP Muriplan treatment planning system 23 . Three pegs (5 mm length, 3 mm diameter) were included in the design to hold the film in place (Fig. 1). A second phantom but with the pellet cavity located in the pelvis region was also designed. The overall dimensions of the phantom were 86.9 mm (length), 30.5 mm (width), and 26.3 mm (height). This design allowed simultaneous irradiation of the film and alanine detectors to allow a direct comparison between measurements. All plans were designed so that the beam was incident on the dorsal surface of the phantom, passing through the film and then through the alanine pellet located directly beneath the film. Gafchromic EBT3 film is thin (< 0.3 mm) and relatively water/tissue equivalent, so is unlikely to perturb the dose distribution downstream of the film. To avoid any risk of the bulkier alanine pellets perturbing the dose to the film, the alanine pellets were located beneath the film. The use of a realistic murine phantom over a simple cylindrical geometry allows a representative test of the treatment pathway, from phantom positioning, image-guided treatment planning and treatment delivery. The 3D printing parameters and gcode for the final phantom design were prepared using Ultimaker Cura software 24 , and the phantoms were printed using an Ultimaker 3 (Ultimaker BV, Utrecht, Netherlands) fused deposition modelling (FDM) 3D printer. The material used was polylactic acid (PLA, RS Components Ltd.) which has a density of 1.25 g/cm 3 in raw filament form. Although the phantoms were printed at 100% in-fill, the measured density of the final printed phantoms (1.19 g/cm 3 ) was lower than that of the raw material due to effects Figure 1. The 3D printed murine dosimetry phantom. An Ultimaker 3 fused deposition modelling printer was used to create bespoke phantoms capable of securing Gafchromic EBT3 film over an alanine pellet to capture an absolute dose measurement and dose distribution with reference to a national primary standard. The assembled phantom is shown on the right. A second version of the same phantom was printed with the alanine pellet cavity situated in the pelvis region. within the printing process 22 . The choice of PLA for the phantom material was made based on its availability and suitability as a tissue substitute 25,26 . The lungs were incorporated into the model as air cavities. The geometrical fidelity of the printed phantoms was assessed in terms of the distance-to-agreement (DTA) between the surface of the phantom (imaged by CBCT) and the design (defined by STL file). The resulting distribution of DTA values was centred on zero with a standard deviation of 0.3 mm, showing that the final printed phantoms accurately replicated the design. Audit procedure. For logistical reasons one institution who took part in the questionnaire was unable to participate in the dosimetry audit. Two murine phantoms (for separate brain and pelvis irradiations) were delivered to the remaining six institutions, herein referred to as S1-S6. Each of these institutions had their own SARRP installation. The audit workflow ( Fig. 2) was designed to simulate the procedure of a typical in vivo experiment: CBCT acquisition, treatment planning and beam delivery. For the end-to-end audit, 2 scenarios were assessed: 1. Delivery of a simple static field to the pelvis using a 10 × 10 mm 2 collimator, with the bed and gantry at 0°. 2. Delivery of a more complex 90° arc field to the brain using a 5 × 5 mm 2 collimator. The bed and gantry angles were 90° and − 45° to 45° respectively, to create an arc in the sagittal plane intended to be representative of techniques designed to spare one hemisphere of the brain. The arc plan was intentionally designed to be more challenging than the static field plan, and to test the limits of what might be attempted in a pre-clinical setting were a field size to be chosen to conform tightly to the intended target. The plan layouts are illustrated in Fig. 3. For both scenarios the prescribed dose of X-rays was 12 Gy to the isocentre, set in the centre of the alanine detector, and the source settings were 220 kVp and 13 mA. Each scenario was planned and delivered twice, with new alanine and film, to obtain repeat measurements. The prescribed dose of 12 Gy was chosen to suit the sensitivity range of the alanine dosimeters. Each centre was provided with a protocol documenting the procedure and equipment handling instructions (Supplementary Table S1), four pre-labelled laser cut films and alanine pellets and spares. For the SARRP irradiations, participants were instructed to follow their standard operating procedure to acquire a CBCT of the phantom with the alanine and film in place and develop a treatment plan on the resulting image using the SARRP Muriplan TPS. As per standard SARRP operating procedure, soft tissue and lung (air) were segmented using pixel intensity thresholding and the standard bulk density overrides were applied. Alanine dosimetry. The alanine detectors were purchased from the alanine dosimeter reference service at the National Physical Laboratory (NPL) to measure the absorbed dose to water 27 . The detectors were 5 mm in diameter and 2.3 mm in height. After use, the alanine pellets were returned to NPL for readout, < 57 days after irradiation. The dose reported is traceable to the primary standard for Cobalt-60 beam quality and therefore a correction factor is required when used with low and medium energy X-rays due to an energy dependence 23 . This correction factor is based on each irradiator's half-value layer (HVL), the thickness of a material (most often aluminium or copper) required to attenuate the intensity of radiation by half 28 . The HVL value was provided by each institution, either following the manufacturer's specification (S3, 4, 5 and 6) or from their own commissioning measurements (S1 and 2). The correction factor corresponding to each HVL value was calculated using the formula, described by Silvestre Patallo et al. 23 based on the correction factors measured in the NPL's medium energy reference beam using HVL values between 0.5 and 4 mm Cu. The resulting absorbed doses were compared to the median planned dose to the pellet volume and the percentage difference calculated. To determine the median planned dose to the pellet volume, the CBCT image and dose grid were first exported from the TPS. Both datasets were imported into an Octave environment 29 for analysis, where the centre of the alanine pellet was identified within the CBCT image. A volume of diameter 5 mm and height 2.3 mm (matching the dimensions of the alanine pellet) was generated centred on this location, and was mapped onto www.nature.com/scientificreports/ the dose grid. All voxels within this volume were identified, with the median value being used as the planned dose in the analysis. Gafchromic EBT3 film dosimetry. A set of calibration films was irradiated using the 300 kV source at the NPL to allow a calibration curve to be generated for the batch of Gafchromic EBT3 film used in the audit. Irradiations were delivered using a beam quality of 0.5 mm Cu HVL, source-to-surface (SSD) distance of 75 cm with solid-water (WT1) slabs arranged to provide 2 cm depth and 20 cm backscatter. 5 × 4 cm sections of film were irradiated at doses of 0, 1, 2, 3, 4, 5, 6, 9, 12 and 15 Gy to create the calibration reference. All films (calibration and measurements) were scanned using an Epson 10000XL flatbed scanner in transmission RGB mode, colour corrections disabled, at a spatial resolution of 400 dpi and a bit depth of 16-bits per colour channel. A minimum time between irradiation and scanning of 24 h was left to allow the polymerisation of the active layer within the film to stabilise 9,30 . Scanned images were stored in TIFF format and all subsequent analyses were done using the original spatial resolution of 400 dpi. All images acquired using the Epson 10000XL scanner were pre-processed prior to use to correct for scanner non-uniformity issues. The correction factor at each pixel of a scanned image is dependent on three things: (1) the colour channel; (2) the position of the pixel on the scanner bed (in terms of the distance from the central axis of the scanner); (3) the darkness of the film at that point. A correction map was previously created using a series of films, uniformly irradiated at doses from 0 to 25 Gy, and scanned at all positions across the scanner bed. This allowed a correction map to be created as a function of colour channel, position and optical density. The calibration curve was parametrised using an equation of the form shown in Eq. (1), as described by Aitkenhead et al. 31 . where A, B, C are the fit parameters and O.D. represents the optical density. Figure 4 shows the calibration films, the fits and residual differences between the fits and the delivered doses for each colour channel. To fit the exact geometry of the 3D printed design, sheets of Gafchromic EBT3 film were laser cut to form the outline of the phantom and accommodate the pegs. To generate the dimensions for the laser cutting, Netfabb was used to convert the model into slices and the slice corresponding to the film location was exported as a DXF file, compatible with the laser cutting software. The asymmetric geometry of the film aided both the fitting of the film into the phantom in the correct orientation, and locating the position of the film in the treatment plan during its analysis. Film measurements of the delivered dose were compared to that calculated by the TPS using gamma analysis 32 . Analysis was performed within an Octave environment 29 using in-house software which has previously been clinically commissioned for other applications 31,33,34 . For each centre, the film data formed the reference dataset while the planned dose grid formed the evaluation dataset, following the terminology used by Low et al. 35 . The same 3D planned dose grid exported from the TPS as part of the alanine analysis was as the evaluation dataset. The software performed a full 3D gamma analysis for the reasons described by Pulliam et al. 36 , calculating the www.nature.com/scientificreports/ gamma index for each pixel in the 2D reference image by minimising the dose difference (DD) and DTA within the 3D evaluation image. The film analysis procedure was designed to allow separate evaluation of (1) the positional agreement between the planned and delivered dose, (2) the absolute dose delivered and (3) the shape of the dose distribution: 1. For each film, the outline geometry and 3 holes corresponding to the pegs shown in Fig. 1 were used to geometrically locate the film within the phantom. The position of the film was then manually adjusted by applying small offsets to best match the field edges to the planned dose distribution. The positional offset applied was recorded as the residual shift. 2. The reference (film) data was normalised to the evaluation (plan) data prior to the gamma analysis in order to assess differences in the dose distribution shape without results being dominated by any difference in absolute dose. Normalisation factors were calculated for all pixels in the film within the 90% isodose region (relative to the prescribed dose of 12 Gy) relative to the corresponding point in the planned dose grid. The median of these factors was taken as the overall normalisation factor for that film. We chose to use the 90% isodose area for normalisation in order to include only the high dose region, excluding the penumbral region where agreement between film and plan is less likely to be reliable. 3. The shape of the dose distribution was evaluated by performing gamma analyses for a range of DD criteria from 2 to 7%, and a DTA criterion of 0.3 mm. Dose differences were evaluated relative to the local dose. All pixels in the film corresponding to a dose > 4% of the prescribed dose (12 Gy) were evaluated within the analysis. The normalisation factor described above also provided a means to evaluate the agreement between the film and plan in terms of absolute dose. A normalisation factor equal to 1 would imply that the film measurement agreed in terms of absolute dose with the TPS (on average, over the area within the 90% isodose line), while a normalisation factor greater than 1 would correspond to the TPS dose being greater than the film. This method of evaluating the absolute dose over an area is more reliable than evaluating the dose to a single point, as it is insensitive to noise in the film measurement (Note that we use the term absolute dosimetry to refer to measurement of the magnitude of the delivered dose, to distinguish from the measurement of the shape of the dose distribution). Results Questionnaire feedback. A summary of the questionnaire results can be found as Supplementary Table 2 online. All institutions agreed that audits and defined dosimetry protocols are important. Most participants suggested that a dose tolerance of within 5% was an acceptable dose agreement and one centre proposed that < 15% would be satisfactory. Variations between centres arose with the questions pertaining to QA tests. Using various ionisation chambers these were either completed every 6 months (in two centres by the manufacturer), every 2 months, whenever the machine was used or daily. The calibration of three of these farmer-type ionisation chambers could be traced to the national primary standard and another two centres' chambers were calibrated by the chamber manufacturer (PTW-Freiburg GmbH). Output checks were performed either annually, bi-annually, monthly or bi-monthly and the output was calibrated at least every 2 years. With regards to reporting parameters, each institution gave the following responses: • S1-No response • S2-Device used, gating used, dose delivered, dose rate, fractionation, image guidance, irradiation technique, field size, SSD, backscatter, couch position, D95, HVL, voltage, filtration, dosimetry protocol (air or water), output measurements, depth, backscatter, target medium, calibration conditions. • S3-No response Although the suggestions varied widely, all participants agreed that more informed reporting would lead to better reproducibility of research. Alanine dosimetry. The HVL thickness of the individual irradiators can be used to account for the difference in response of alanine between medium energy X-ray and Cobalt-60 beams 23 Table S3). The static beam alanine measurement, in which the 10 × 10 mm 2 field covered the whole alanine pellet in the plan (5 mm diameter), was used as an accurate determinant of the delivered dose. The results, after applying the energy dependence correction factors, are presented in Fig. 5a. All but 1 measurement achieved < 10% deviation from the planned dose, and more than half of the measurements were within ± 5%. The mean absolute difference was 4.8%. The difference between the planned dose and alanine measurement across all centres was 4.1 ± 4.3% (mean ± standard deviation), with an overall range of − 2.3 to 10.5%. Alanine irradiation in the phantom using the more complex beam configuration of an arc (Fig. 5a) resulted in lower measured doses than for the static beam. All but 1 measurement achieved < 10% deviation from the planned dose, and more than half of the measurements were within ± 5%. The mean absolute difference was 4.5%. The difference between the planned dose and alanine measurement across all centres was -1.2% ± 6.1% (mean ± standard deviation), with an overall range of − 13.1 to 7.7%. For each individual SARRP, the dose difference between the repeat measurements ranged from 0.1 to 1.0% for the static beam, and from 2.0 to 3.7% for the arc delivery. A summary of the alanine dosimetry results can be found in Supplementary Table S3. Gafchromic EBT3 film dosimetry. The residual shifts required to match the film position to the plan were consistent across all centres, with an average residual of 0.68 ± 0.35 mm (mean ± standard deviation) across all centres and deliveries. These residuals may represent the total of all sources of error, including delivery issues such as the difference between the true isocentre position and the isocentre position modelled in the TPS, as well as measurement issues such as the accuracy of positioning the film within the phantom. For Gafchromic EBT3 film, the absolute dose measurements were derived from the normalisation factors computed during the gamma analysis procedure (described earlier), and the results are presented in Fig. 5b. 79% of measurements achieved < 10% deviation from the planned dose, and more than half of the measurements were within ± 5%. The mean absolute difference from the planned dose was 6.7% and 4.2% for the static and arc deliveries, respectively. The difference between the planned dose and film measurement across all centres was 6.0 ± 6.2% (mean ± standard deviation) with an overall range of − 3.2 to 14.7% for the static deliveries, and 1.3 ± 5.2% (mean ± standard deviation) with an overall range of − 9.9 to 10.7% for the arc deliveries. www.nature.com/scientificreports/ Figure 6a illustrates the results of the gamma analyses for one film from each centre for the static plans. Three key observations may be made: 1 For several centres (e.g. S1, S2, S5 and S6), the measured dose outside the field was notably higher than the planned dose, as can be seen by the red regions surrounding the field in the gamma images. 2 The largest discrepancy between the planned and delivered dose distributions is in the corners of the square fields, shown by the blue regions in the corner of each field in the gamma images for centres S1-S5. In the measured fields the corners have a more rounded profile and a lower dose than predicted by the TPS. 3 At centre S6, the 10 × 10 mm 2 field was in fact modelled by a 10 mm diameter circular field. This was a deliberate choice by that centre to minimise the number of apertures that had to be commissioned. As noted above, the corners of the square field measured in the film disagreed with those modelled by the TPS, therefore the use of a circular model is not as unrealistic as might be expected. Figure 6b illustrates the results of the gamma analyses for one film from each centre for the arc plans. The results for all gamma analyses for the static and arc plans are summarised in Fig. 7. Although the 10 × 10 mm 2 static plans were simpler than the arc plans, the static films consistently had a smaller proportion of passing pixels (γ ≤ 1) mainly because the results were dominated by the discrepancies at the corners of the field and in the out-of-field region. In contrast, in the arc plans the steeper out-of-plane dose gradient tended to result in a higher proportion of passing pixels. Results for all centres were broadly comparable in terms of the trend of the gamma pass rate as a function of the dose difference (DD) criterion. A summary of the EBT3 dosimetry results can be found in Supplementary Table S4. Figure 8 compares the absolute dose measurements obtained using Gafchromic EBT3 film and alanine pellets. Data is only shown for matched film and alanine measurements: i.e. where the film and alanine were irradiated together on a single irradiation to allow for a fair comparison of EBT3 and alanine results. No matched film and alanine measurements were available for the arc plan at centre S4 due to the alanine pellets being damaged following irradiation and prior to read-out. The data in Fig. 8 is presented in terms of the ratio of the planned dose to the measured dose, showing results for alanine and EBT3 on the x and y axes, respectively. The diagonal line indicates the line of agreement between alanine and EBT3. For data lying to the upper-left of the diagonal line the EBT3 measurement was lower than alanine, while for data lying to the lower-right of the diagonal line the EBT3 measurement was greater than alanine. Figure 8a compares the EBT3 and alanine measurements of dose for the static irradiations. The data lie close to the diagonal, indicating that EBT3 and alanine were generally in good agreement. However, the film measurements were typically higher than alanine (by 1.6 ± 2.2%, mean ± standard deviation), and therefore the data are found mainly to the lower-right of the diagonal. Figure 8b compares the EBT3 and alanine measurements of dose for the arc irradiations. The agreement between film and alanine was worse than for the static plans, which was likely due to the increased complexity of the arc plans, with steeper dose gradients and a high dose region no larger than the dimensions of the alanine pellet. Again the film measurement was typically greater than the alanine, although with a larger variation (2.9 ± 5.6%, mean ± standard deviation) than for the static plans. Comparison of alanine and Gafchromic EBT3 film measurements of absolute dose. Practical problems encountered during the audit. During the audit several problems were encountered: 1. One of the phantoms had warped during printing making the alanine cavities smaller than designed, and this was not detected during phantom quality control. This meant several pellets were damaged upon removal and were therefore unreadable (arc data for S4). The poor fit of the alanine pellet in the warped phantom also resulted in an air gap beneath the film. This resulted in an artefact in the planned dose distribution adjacent to and downstream of the film. To prevent this affecting the gamma analyses, the planned dose in the air gap was corrected using interpolation. Although PLA is more robust than other commonly used 3D printing materials, such as acrylonitrile butadiene styrene 25 , for future phantom designs the size of the pellet cavity should be reviewed to ensure both potential warping of the phantom and variation in the dimensions of individual alanine pellets are taken into account. 2. One centre mislabelled the plan information upon return so these had to be manually matched to the films during analysis. 3. One centre (S4) mis-interpreted the planning guidance and used an arc in the transverse plane rather than one in the sagittal plane. However, for the purpose of the audit the use of a transverse arc was also acceptable, since the aim was to test agreement between the planned and delivered dose. However, this demonstrates the need for unambiguous guidance in multi-institutional studies. 4. One centre inserted the film the wrong way in the phantom, but due to the symmetrical structure of EBT3 film this had no impact on the analysis. 5. Finally, one centre sent incomplete dose information and the original plan was removed from their TPS. However, it was possible to re-create the plan within Muriplan using the data that had been exported, allowing the dose to be recalculated. www.nature.com/scientificreports/ Discussion We present results from a preclinical dosimetry audit of Xstrahl SARRP systems at 6 institutions in the UK using a realistic murine phantom. The results of the questionnaire suggest that without a routine protocol there is still some way to go before a consensus is reached across the country. One of the contributing factors to a paucity of rigorous dosimetry protocols is insufficient dosimetry knowledge or support from clinical physics colleagues 4 . Some centres rely on the manufacturer to complete the dosimetry checks, which has the advantage that the checks are done in a consistent fashion by staff who are expertly trained on the system. However, this is not a guarantee that the irradiator has been properly calibrated 37 , and further checks should be performed by appropriate personnel for validation. The responses to the questionnaire also indicate that there may be a lack of support at the institutional level, and the lack of independence in the QA process increases the risk of a systemic problem going unnoticed 37 . Also of importance is the wide variation in the frequency of output checks (bi-monthly to annually) and the traceability of the chambers used for those checks. These should be addressed as the output of a SARRP system can potentially drift over time, with Feddersen et al. 10 reporting a decline of almost 4% over an 18 month period. There did not appear to be any correlation between the frequency and traceability of the QA checks performed and the accuracy of the dosimetry within the present study. www.nature.com/scientificreports/ Only one centre reported most of the suggested requirements for accurate reporting recommended by Verhaegen et al. 18 . Similarly, many peer-reviewed articles fail to report basic details that are necessary to allow a study to be reproduced or compared to other studies 7,19 . Incomplete reporting of these experimental parameters adds to the biological sources of error that are complex and poorly understood, and is often attributed to insufficient physics expertise among users 19 . One simple proposed standardized QA methodology is to make use of the in-built electronic portal imaging device, which has shown to be a stable and convenient tool to assess beam quality, energy, output, profile and targeting and verify delivered doses 38,39 . A thorough QA procedure would include the use of ionisation chambers for calibration (with reference to a primary dosimetry standard), film for 2D measurements and a smaller detector to validate dose at submillimetre resolution 40,41 , such as MOSFETs (metal-oxide semiconductor field-effect transistors) or TLDs (thermoluminescent detectors) 4 . However to implement this the issue of physics expertise must first be addressed. Currently, most centres included in this audit use ionisation chambers for QA. Only two centres also use film as part of their QA process. When the SARRP was first developed over a decade ago the suggested in vivo dosimetry tolerance was 5-10% 2 , consistent with the then 5% target used by audits of clinical low or medium energy X-ray irradiators, with action points if results exceed 10% 42 . This was especially important since it is documented that a 10% dose difference can lead to mortality rates in some mice strains of up to 90% 43 . Apart from 1 delivery from centre S6, the static field deliveries were all within 10% of the planned dose as measured using the alanine dosimeters, with the majority being < 5% (see Fig. 5a). The delivery which failed to meet the 10% tolerance was marginal (being 10.5% high). There are several uncertainties that may contribute to these discrepancies. The accuracy of the HVL measurement and the calculation of the correction factors for the alanine energy dependence (estimated to be 4.8%) or the difference in beam quality between the reference beam at NPL ( 60 Co) and the SARRPs (X-ray) used 23 . Additional uncertainties in the applied correction factors may come from the difference in the spectra between the SARRPs and the NPL's reference beam 23 . Using the nominal HVL thickness of 0.67 mm Cu instead of 0.847 mm Cu, which is due to additional beam gating equipment, increases the dose difference by 2.4%. Furthermore, the signal readout has been known to degrade over time, especially in humid environments 44 . However, here the maximum time between irradiation and readout being < 2 months and the pellets being stored in two sealed envelopes the signal should have remained stable 44 . There may also be contributing uncertainties that are related to the TPS calculations such as segmentation thresholds, commissioning or targeting, which are out of the scope of this investigation. It was assumed the CBCT dose was negligible (< 0.85 cGy) in line with other studies 2,45 , and did not contribute to the delivered dose difference. Apart from 1 delivery from centre S5, the alanine measurements were also all within 10% of the planned dose for the arc deliveries (see Fig. 5a). That the measured doses were lower than for the static beams could be due to the width of the arc field and pellet diameter being the same. A small error in the isocentre targeting, either from user or TPS inaccuracies, would result in incomplete coverage of the alanine dosimeter and therefore the average dose measured over its volume would be reduced. There are additional uncertainties due to the rotation of the bed (during the CBCT acquisition) and gantry (during radiation delivery), which may contribute to the overall targeting uncertainty. Gafchromic EBT3 film is well established as a tool for absolute dosimetry from 0.01 to 30 Gy 30 . The use of the red colour channel alone allows for accurate dosimetry up to 8 Gy, while the use of multiple colour channels extends the range and reduces the uncertainty in the measured dose 46 . In this study we used the red and green channels for dosimetry up to 15 Gy. Film provides the ability to assess certain features of the delivered dose distribution that cannot be evaluated using point dosimeters such as alanine pellets: the shape of the dose distribution in 2D; the position of the delivered dose; and the dose deposited outside the high dose region. Each of these types of error has the potential to lead to inaccurate conclusions being drawn from in vivo experiments 47 . Increased focus on validating these aspects of the delivered dose distribution, rather than focussing only on the absolute dose delivered to the target region, would help to refine experiments in several ways: better agreement between planned and delivered dose distributions would increase confidence in delivered doses, consequently reducing the number of animals required, and improving control of the delivered dose away from the target may help to reduce the radiationinduced side effects experienced by the animals. The results illustrated in Fig. 6a show that agreement between the plan and EBT3 film measurements was generally poor at the field corners, where the film measurements showed a more rounded profile and lower dose than predicted by the TPS, and in the out-of-field regions, where the film measurements were notably higher than the planned dose. Dose errors in these regions could be a concern for experiments where the dose to normal tissues in close proximity to the target is important. These issues were observed in all 10 × 10 cm 2 static field measurements for all centres, indicating that they are not due to a delivery error. Further investigation is warranted into the exact cause of these discrepancies. Film dosimetry is least reliable at low dose levels, and therefore the out-of-field dose discrepancy could potentially be explained by the limitations of film dosimetry. This is less likely for the discrepancies seen at the corners of the field, which are not in a low dose region and spatially are well within the resolution limits of film dosimetry. Previous studies have suggested that the superposition-convolution dose calculation algorithm used in Muriplan does not accurately model the penumbra 9,39 . Implementation and evaluation of alternative dose calculation algorithms, such as using a Monte-Carlo approach, is worthy of investigation since they may have different behaviour in the out-of-field regions. Other parts of the planning and delivery process may also benefit from investigation, such as the use of bulk density overrides to segment tissue types within the CT image. It is worth noting that similar discrepancies were not observed in the arc dose distributions (Fig. 6b), perhaps because of the use of a smaller field size, and because the relative motion of the beam tends to soften field edges parallel to the axis of rotation. In relation to the use of Gafchromic EBT3 film for measurement of absolute dose, it should be noted that our calibration films were irradiated using a 300 kVp beam, while the SARRP systems tested within the audit www.nature.com/scientificreports/ used a 220 kVp beam. The response of EBT3 film is known to be sensitive to energy when used for kilovoltage beams 46 , although the formulation of the active layer in the film has been designed to minimise this 48 . Results reported by Bekerat et al. 48 show that the energy dependence is small at energies above 70 keV, and within the range 220-300 keV is likely to be within a few percent at most. Nevertheless, this has the potential to introduce a small systematic error to the EBT3 absolute dose measurements within the audit. It was observed that the EBT3 measurements of dose were typically higher than alanine, by 1.6 ± 2.2% for the static plans and 2.9 ± 5.6% for the arc plans. It is worth highlighting that this difference was greater for the arc plans, which is consistent with our earlier remark that the alanine dosimeters may not receive the full planned dose in the event of a small set-up error, due to the width of the arc field and pellet diameter being the same. However, we found no relationship between the dose difference and the residual positional shifts obtained during the film analysis procedure. The density of soft tissues typically ranges from 0.95 g/cm 3 (for adipose tissue) to 1.05 g/cm 3 (for muscle) 49,50 . Furthermore, the phantom used during the Muriplan TPS commissioning is kV-equivalent solid water 11 . Keeping in line with published dosimetry protocols, dosimetry phantoms should be a density close to water (1 g/cm 3 ) such that the measurements obtained are within a few percent 51,52 .The ICRU report 44 states corrections factors may be required for absorbed dose measurements obtained with phantoms that introduce uncertainties greater than 1% 50 . The density of the phantoms used in the audit was 1.19 g/cm 3 . This difference potentially impacts on the accuracy of the dose calculation. Within the TPS, the tissue segmentation allows voxels within an image to be assigned as one of 5 discrete materials (air, lung, fat, tissue or bone) whose densities are defined according to ICRU report 44 50,53 . An underestimate in the density of the material defined as 'tissue' will lead to the TPS underestimating the attenuation of the beam within the phantom, which we estimate could lead to an error of 1-2% in the calculated dose at the film or alanine detectors. The magnitude of the dose error will be dependent on the field size, the geometry of the phantom, the depth of the dosimeter, and the SSD. This may contribute to the behaviour observed in Fig. 5, where the dose differences between measurement (both alanine and film) and plan are consistently higher for the static plans than for the arc plans. In terms of the variation between institutions any dose error due to the density of the phantom would be systematic, having the same impact on all centres. For future phantom or audit work the choice and density of material used for the phantom should be carefully specified and checked to improve the accuracy of absolute dose measurements. Finally, many of the practical issues encountered during the audit arose due to the logistics of it being a postal audit as opposed to having being undertaken by the authors. However, a benefit of a postal audit is that it provides information regarding the real use of the machines by the end users. With the exception of the phantom warping, the issues had no real impact on the results but highlight the need for clear instruction for future postal audits. Conclusions Regular end-to-end dosimetry audits complement the QA performed by the user, testing all stages of the planning and delivery process, and provide confirmation that centres' practices and results are consistent with the wider community. This audit shows the potential of using realistic phantom geometries for evaluation of dose distributions that are representative of experimental scenarios. The use of two different types of dosimeter (film and alanine) allows different features of the dose distributions to be evaluated, and also provides the means to check consistency between the different dosimeters, in this case in terms of absolute dose. This preclinical dosimetry audit found the delivered doses for the simple static field plans to the pelvis to be within 10% of the planned dose to the isocentre for all but one of the measurements across the six centres, but with an overall range of 12.3% between the lowest and highest measured dose, and an overall mean absolute difference of 4.6%. For the more challenging arc plans to the brain all but one measurement was within 10% of the planned dose and the overall mean absolute difference was again 4.6%, but the overall range between the lowest and highest measured dose was larger, at 21.4%. Standardisation of dosimetry protocols would be desirable to improve the agreement between centres in order to prevent such dose differences leading to significantly different biological responses in in vivo experiments. We recommend that phantoms such as the one reported in this study be adopted into routine dosimetry QA protocols. Consistent and regular checks will ensure accurate and precise dosimetry and improve the reproducibility of research results at different institutions.	2023-02-22T14:55:13.061Z	2022-04-26T00:00:00.000	{ "year": 2022, "sha1": "053f2e4a528e9a25325fd2259faa8255d25709b8", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41598-022-10895-5.pdf", "oa_status": "GOLD", "pdf_src": "SpringerNature", "pdf_hash": "053f2e4a528e9a25325fd2259faa8255d25709b8", "s2fieldsofstudy": [ "Medicine", "Physics" ], "extfieldsofstudy": [] }
259246071	pes2o/s2orc	v3-fos-license	The combined application of bleomycin and triamcinolone for the treatment of keloids and hypertrophic scars: An effective therapy for treating refractory keloids and hypertrophic scars Abstract Background Keloids and hypertrophic scars frustrate patients by the deformity of appearance and organ dysfunction. Many modalities had been tried in clinic practice, but the results are unsatisfied. Objective We retrospectively analysed the combined application of bleomycin and triamcinolone for the treatment of keloids and hypertrophic scars. Methods The combination of bleomycin and triamcin acetonide was applied to the treatment of keloids and hypertrophic scars, 86 cases accepted the treatment. Follow‐up 2–5 years after treatment. Results (1) The pain of scars and itching symptoms have basically subsided through treatment. (2) After drug injection treatment, the keloid began to shrink, some of the keloids disappeared. (3) Small keloids did not recur after treatment. Large keloids had local recurrence after treatment. When further treatment was given, the recurrence disappeared. Conclusion The combined application of bleomycin and triamcin acetonide can effectively cure keloids and hypertrophic scars. LUO basically disappeared, the hyperplasia of keloids was effectively suppressed, and the volume of keloids shrank to varying degrees or completely subsided. Below we conduct a retrospective analysis of past cases. MATERIALS AND METHODS Cases: From June 2012 to August 2020, a total of 86 patients with keloids were treated. Age from 7 to 72 years. There were 35 males and 51 females. Location of keloids distribution and size: 18 cases of mandibular margin and masseter muscle area, 21 cases of ear, 23 cases of chest, 13 cases of shoulder, eight cases of back, three cases of perineum. Scar size: from 0.6 cm × 0.6 cm to a huge keloid on the chest and back and perineum. Area 0-16 cm 2 47 cases, 16-49 cm 2 25 cases, 49 cm 2 or more 14 cases. Injection method: Inject the drug into the scar with a 1-mL syringe. Bleomycin is used in 0.05 million units/cm 2 and triamcinolone acetonide is 5 mg/cm 2 . When injecting the drug, the base of the keloid should be covered so that bleomycin can effectively destroy the blood supply from the base of the keloid to inhibit the growth of the keloid and promote its atrophy. The depth of administration of bleomycin is not too shallow, and the drug concentration is not too thick, otherwise it is easy to cause keloid ulceration. The concentration of triamcinolone acetonide is not too thick; otherwise it is easy to cause uneven regression of scars and bumps on the surface. When injecting drugs, the drug is injected in a radial dispersal manner, and the spacing between the drug injections is less than 0.3 cm, so that the drug is evenly distributed in keloids as much as possible. Interval of administration: The interval of treatment is determined according to the state of scar hyperplasia and the size of the scar. If the scar is dark red, full, and shiny on the surface, the scar is relatively active, and the interval between dosing is 2 weeks. If the scar is slightly dark red and has a skin texture on the surface, it can be spaced for 4 weeks. Smaller keloids are generally spaced 4 weeks apart. Due to the limitation of the dosage of the drug, the treatment of larger keloids cannot completely cover all the keloids in one injection. The drug can be taken for 3 consecutive days so that the keloids can be completely covered, and the treatment can be injected again after 3-4 weeks. Generally, the interval between initial sessions is short, and the interval between later stages is long. When skin wrinkles appear on the surface of the keloid, the interval between dosing is 1 month. The interval between doses is not fixed and depends on the change of the scar lump after each treatment. Clinical indications for stopping keloid drug injection therapy: (1) the clinical symptoms of pain and itching disappeared; (2) keloid color changed: from light red, dark red to close to normal skin; (3) the texture of the scar pimple changed: from the original hardness to softness; (4) the keloid atrophied, and the surface changed from the original hardness and smoothness to skin wrinkles. Keloids shrinked to varying degrees, and some keloids even completely subsided. Surgical treatment of keloid scars Most surgical methods were performed using keloid removal and scar tissue flap repair. After the operation, the tissue healed well, the skin at the original scar area was flat, and the skin color slowly recovered, close to normal skin tone. Laser treatment of keloids After drug treatment, the scar lump did not completely subside, the surface was uneven, and after the keloid subsided, the surface was rough and not smooth, and the carbon dioxide laser grinding treatment was carried out. Laser treatment mode was performed in continuous or dot matrix mode. The timing of laser treatment for keloids was also chosen when the keloids were in a stable state after treatment. After laser treatment at most lesions, the skin returned to near normal skin. A small number of uneven lesions appeared hyperplastic after laser treatment. This phenomenon mainly occured in continuous laser excision of protruding scars, and lasers caused excessive skin lesions. This phenomenon did not occur in dot matrix laser mode. Keloid treatment flowchart: Step I Drug therapy, to make active hypertrophic scars and keloids into stable scars. After medication, some of the scars subsided and flattened, no further treatment was required. Scars with morphological deformities are further treated. Step II Laser or surgical treatment. Morphological repair of scars were done in the stable phase. Scars with uneven surfaces were treated with lasers. Keloids and scars that protruded from the skin were treated surgically. Distribution of cases receiving different treatments at each stage: Drug therapy Laser therapy Surgical operation Step I 86 Step II 21 27 Changes in clinical symptoms The pain of scars and itching symptoms have basically subsided through treatment, and the rate of regression was different. Mainly related to the size of the scar lumps, the lesions were smaller, the clinical symptoms subsided quickly, and they subsided after 1-2 injections. The lesion was larger and the clinical symptoms subsided slowly. This difference was mainly related to the limited dose of each drug use, the large area of the scar lump, and the fact that a single injection did not completely cover the diseased tissue.The time when the pain and itching of the scar lumps subside and the difference in various size scars were listed in Table 1. Keloid atrophy changes The degree of atrophy was observed when a stable state was achieved after treatment. After drug injection treatment, the degree of keloid atrophy was correlated with the height of keloid hyperplasia and bulge. Keloids with small bulges receded quickly, and keloids with high lift heights receded more slowly. Hyperplastic, highly raised scars did not resolve completely, but will atrophied to varying degrees. Scars with a thickness of ≤4 mm can basically be flattened. The thickness of 5-to 7-mm scars and pimples subsided to varying degrees, some faded more, some faded less; Scars > 7 mm thick were difficult to resolve, and even if they did, the surface was uneven.The TA B L E 2 The degree of keloid regression with different scar thickness was shown. Keloids with small bulges receded quickly, and keloids with high lift heights receded more slowly. Scars less than 4-mm thick are flattened after treatment, scars with a thickness greater than 7 mm are difficult to flatten after treatment. Recurrence of scars The follow-up period after keloid drug treatment was 2 years. Recurrence was judged by resurgence. The recurrence rate of keloids was related to the area of the keloids and the thickness of the hyperplasia. Keloids with small areas were treated with adequate courses of treatment and basically did not recur. Large, thicker keloids that have been treated for a longer period of time still recur, and recurrence was manifested by hyperplasia of local keloids rather than hyperplastic hypertrophy of overall keloids. The recurrent site was again given drug injection therapy and continued to be treated until it was stable. The recurrence was observed for 1 year, and there was still a local recurrence in cases with an area of >49 cm 2 . When counting cases, all cases that have had relapse are classified as relapsed cases. The keloid recurrence with different size of scar area and thickness is listed in Table 3. The preoperation and postoperation of the lesion appearance are shown in Figures 1-3. F I G U R E 1 The appearance of the left ear keloid before treatment was shown. After treatment with bleomycin and triamcinolone acetonide, the erythema of keloid regressed, the keloid changed from hard to soft, and the surface of the keloid became wrinkled. The ear reconstruction was performed, the shape of the outer ear has basically returned to normal. Over the next 2 years of follow-up, the scar remained stable and did not recur. F I G U R E 2 The ear scar keloid looks cherry red before treatment. Then the keloid was removed, and the auricle was repaired with scar flaps after bleomycin and triamcinolone acetonide treatment. The ears returned to normal. In the following 2 years, the shape of the outer ear did not change, and the scar did not recur. F I G U R E 3 The keloid of right scapular and chest before treatment were visibly raised and plump. After bleomycin and triamcinolone acetonide treatment, the scapular keloid became flat and soft, the chest keloid disappeared. Over the next 2 years of follow-up, the scar remained stable and did not recur. DISCUSSION Keloids and hyperplastic scar are difficult to treat, the most important thing is its recurrence, sometimes the recurrence of keloids is even larger than the original keloids, scars will invade the surrounding normal tissues. Current conventional treatments are difficult to work. The treatment of keloids and hyperplastic scar include surgical treatment, hormone injection therapy, radiation therapy, laser treatment, cryotherapy, etc. Standard keloid treatment is surgery after hormone injections or followed by local radiation therapy. According to relevant clinical reports, the recurrence rate of scars and pimples is still high with this treatment method. 1 We take the first drug injection treatment clinically (using bleomycin and triamcin acetonide) and carry out the necessary surgical treatment In our clinical treatment, there is one thing to note: the course of drug treatment should be sufficient; otherwise it is easy to recur. Of course, the recurrence of keloids may also be related to the biological characteristics of each keloid itself, and perhaps some keloid cells are more active, and their own microenvironment is more conducive to keloid recurrence. Next, we intend to improve the method of administration so that the drug can be better evenly distributed within the scar tissue. CONCLUSION The combined application of bleomycin and triamcin acetonide can effectively eliminate the clinical symptoms of itching and pain of scars, make the scars atrophy and flatten, not recur. After treatment, small and medium-sized scars no longer recur; large-area keloids are only local recurrences even if they recur. CONFLICT OF INTEREST STATEMENT The authors have no conflicts of interest to disclose. FUNDING INFORMATION The authors received no specific funding for this work. DATA AVAILABILITY STATEMENT The data that support the findings of this study are available from the corresponding author upon reasonable request. ETHICS STATEMENT The study plan and consent to participate were reviewed and approved by the ethics committee of Peking University, School and Hospital of Stomatology (in 2012, it was implemented as a filing system, but there was no serial number). All patients signed an informed consent form prior to treatment and agreed to use clinical data for academic communication and publication.	2023-06-26T13:01:44.591Z	2023-06-01T00:00:00.000	{ "year": 2023, "sha1": "7457b0516278a2cfa97927d71a8026a93c5a8dd1", "oa_license": "CCBYNCND", "oa_url": "https://doi.org/10.1111/srt.13389", "oa_status": "HYBRID", "pdf_src": "Wiley", "pdf_hash": "7457b0516278a2cfa97927d71a8026a93c5a8dd1", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
257711679	pes2o/s2orc	v3-fos-license	The Association of Combined Per- and Polyfluoroalkyl Substances and Metals with Allostatic Load Using Bayesian Kernel Machine Regression Background/Objective: This study aimed to investigate the effect of exposure to per- and polyfluoroalkyl substances (PFAS), a class of organic compounds utilized in commercial and industrial applications, on allostatic load (AL), a measure of chronic stress. PFAS, such as perfluorodecanoic acid (PFDE), perfluorononanoic acid (PFNA), perfluorooctane sulfonic acid (PFOS), perfluoroundecanoic acid (PFUA), perfluorooctanoic acid (PFOA), and perfluorohexane sulfonic acid (PFHS), and metals, such as mercury (Hg), barium (Ba), cadmium (Cd), cobalt (Co), cesium (Cs), molybdenum (Mo), lead (Pb), antimony (Sb), thallium (TI), tungsten (W), and uranium (U) were investigated. This research was performed to explore the effects of combined exposure to PFAS and metals on AL, which may be a disease mediator. Methods: Data from the National Health and Nutrition Examination Survey (NHANES) from 2007 to 2014 were used to conduct this study on persons aged 20 years and older. A cumulative index of 10 biomarkers from the cardiovascular, inflammatory, and metabolic systems was used to calculate AL out of 10. If the overall index was ≥ 3, an individual was considered to be chronically stressed (in a state of AL). In order to assess the dose-response connections between mixtures and outcomes and to limit the effects of multicollinearity and other potential interaction effects between exposures, Bayesian kernel machine regression (BKMR) was used. Results: The most significant positive trend between mixed PFAS and metal exposure and AL was revealed by combined exposure to cesium, molybdenum, PFHS, PFNA, and mercury (posterior inclusion probabilities, PIP = 1, 1, 0.854, 0.824, and 0.807, respectively). Conclusions: Combined exposure to metals and PFAS increases the likelihood of being in a state of AL. Background The totality of exposures people endure throughout their lives and how those exposures affect health have been referred to as the exposome [1]. Although certain environmental exposures might lead to unfavorable health outcomes, little is understood about how these factors interact or synergize to affect the stress response system [2]. This is especially critical to understand when exposure to metals is mixed with exposure to perand polyfluoroalkyl substances (PFAS). The negative consequences of PFAS and mixed metals may be deleterious. They could have a long-term effect on the impacted populations' social, educational, and economic advancement [3,4]. In many environments with high levels of chronic stress, several metals and PFAS co-exist at moderate to high levels. Individuals maintain physiological balance through allostasis, which involves adjusting bodily characteristics to meet environmental requirements. Homeostasis describes health as a state in which all physiological parameters function within non-changing setpoints. Allostasis, on the other hand, states that there are no setpoints and that the demands of the moment will determine the normal values of markers. However, the body adjusts to the higher set point if the impediments persist [5]. When the setpoint is changed, people are said to be in a state of allostatic load. Allostatic load (AL), an index of persistent physiological stress, is the biological consequence of stress. AL depends on the assumption that repetitive activation of the hypothalamic-pituitary-adrenal (HPA) axis affects multiple organ systems [5][6][7][8]. The wear and tear on the body caused by ongoing exposure to stressors can be measured by AL, which combines markers from systems within the human body to form a comprehensive biological stress index. An adult's well-being is negatively impacted by psychosocial stresses, such as poverty, racial inequality, lack of access to resources, and water and food insecurity, which may be combined with environmental factors to increase AL within populations. At the individual and population levels, real-world human exposure to stressors is extraordinarily varied and temporally dynamic. Humans are constantly exposed to intricate chemical combinations of PFAS, metals, and other environmental pollutants [9,10]. Data analytics techniques provide a novel way to analyze the combined risk of various exposures in order to develop methodologies to properly identify and evaluate their impact on indices of stress, such as AL, because we do not fully understand the combinational nature of these exposures [11]. Human Exposure Pathways to PFAS and Metals According to the Agency for Toxic Substances and Disease Registry at the Centers for Disease Control and Prevention (CDC), metals such as cadmium (Cd), arsenic (As), lead (Pb), and mercury (Hg) are among the top 10 most toxic substances. Most people are exposed to metals through ingestion (through water and food), inhalation (through cigarette smoke or industrial products), or skin contact (through paint or soil) [12]. For example, As comes in two forms: the inorganic form is highly toxic, while the organic form is not. Most people are exposed to inorganic As, which is found in soil and groundwater, through drinking water, often from unregulated private wells. Most people are exposed to organic arsenic, which is found in fish and shellfish, through ingestion [13]. In the United States, people of different races, ethnicities, and socioeconomic backgrounds experience widely varying degrees of exposure. For example, non-Hispanic blacks have higher Pb exposure than non-Hispanic whites [14]. Humans most commonly absorb toxic PFAS through their diets [15]. Inhalation of air or dust containing PFAS particles is another route of exposure. Over the past decade, there has been extensive research on the dangers of PFAS exposure for people's health. The CDC, for example, has set limits on PFAS concentrations in drinking water (70 ppt for PFOA and PFOS). PFAS spreads through many sites, including landfills and sites where PFAS has been processed. E-waste sites, for example, leach PFAS into groundwater, soil, and air, while wastewater treatment plants (WWTPs) release PFAS-laden effluent into rivers, lakes, and farms [16]. PFAS from treated or untreated effluent enters sewers, rivers, lakes, and oceans through aquatic ecosystems, making water the ultimate repository of PFAS in the environment [2]. Pregnant and parturient women, elderly people, children, and neonates are the most vulnerable to PFAS exposure, which can cause thyroid, lung, kidney, reproductive organ, metabolic, brain, and behavior disorders, obesity, type 2 diabetes, proteinuria, hematuria, immunosuppression, and adverse pregnancy outcomes [17]. Bayesian Kernel Machine Regression (BKMR): A Mechanism for Monitoring Multiple Environmental Exposures Bobb et al. introduced Bayesian kernel machine regression (BKMR) for analyzing mixtures within the R statistical program [18]. By using the (bkmr) package for the R programming language, BKMR was created to estimate the health effects of pollutant mixtures and is used for toxicological, epidemiological, and other applications. It does this by using procedures from Gaussian predictive methods or hierarchical variable selection [18,19]. The estimation of health outcomes of the mixtures under kernel function is modeled on the exposure variables by adjusting for potential covariates or cofounder factors [20]. These procedures can address the possible collinearity of the mixtures' components and test the exposures' overall health effects [21]. Ultimately, BKMR modeling is a technique that (1) models the exposures and outcomes comprehensively, (2) evaluates the components of chemicals independently of the independent-dependent function, (3) evaluates the effects of mixtures of chemicals, and (4) distinguishes the necessary chemical mixtures for any dataset that is simulated [19,21]. BKMR is also used to solve the challenges encountered when evaluating the health impacts of chemical mixtures (i.e., PFAS and metals). In epidemiological and toxicological studies, BKMR helps solve problems such as collinearity and strong correlations between exposures [22]. BKMR uses variable selection that produces and estimates posterior inclusion probabilities (PIPs) values, which measure the values of variable importance for each exposure in a mixture [18,20]. This study using BKMR hypothesizes that exposure to metals and PFAS is associated with high levels of AL. PFAS and metals were chosen due to the unique opportunity to assess combined exposures to organic and inorganic contaminants, the extensive research on both groups of contaminants with National Health and Nutrition Examination Survey (NHANES) data, and the vast historical and emerging research related to these contaminants. To test this hypothesis, data from the NHANES were used to identify the factors most critical in combined exposures to PFAS and metals. Study Cohort and Design Data from the NHANES 2007-2014 of adults aged 20 years and over were utilized in this investigation. This dataset is a representative sample of non-institutionalized people residing in all 50 U.S. states and the District of Columbia. The U.S. Centers for Disease Control and Prevention (CDC) collected the data, which are available in two-year cycles and include multi-year, stratified, multi-stage, and clustered samples. The population of the United States is represented by the statistics for four cycles within 2007-2014. Selected individuals in the NHANES underwent a physical examination and an interview. The participants' blood was extracted, and samples were sent to a laboratory for evaluation. On the NHANES website of the CDC, additional descriptions and information about the study, as well as the steps and processes involved in data collection, are provided. The association between the various PFAS/metals concentrations and AL levels was examined using weighted data in order to produce sample estimates, which reflect how many people in the U.S. population one individual represents. PFAS and Metals Measurements There were two examination sessions each day. Exams in the morning, afternoon, or evening were randomly assigned to participants. After fasting for nine hours, participants were instructed to consume 75 g of dextrose (10 oz. of glucose solution) within 10 min after the initial blood draw. After the first blood draw was taken, a second blood sample was taken [23]. PFAS Quantification At the mobile examination center (MEC), the CDC gathered blood samples for laboratory analysis to evaluate serum for PFAS. Polypropylene or polyethylene containers were used to store the serum samples. The vials were subsequently shipped to several laboratories across the country. Sample analysis was performed at every survey location under the same conditions, owing to the controlled environments at separate facilities. In order to concentrate the analytes (PFAS) in a solid-phase extraction column, one aliquot of 50 mL of serum was injected into a commercial column switching system after being diluted with formic acid. High-performance liquid chromatography was used to separate the analytes from one another and the other serum constituents. A negative-ion Turbo Ion Spray (TIS) ionization source was utilized for detection and quantification (DOQ). Tandem mass spectrometry was used to change liquid-phase ions into gas-phase ions, utilizing a variation of the electrospray ionization source. These PFAS can be quickly detected in human serum using this technique, with detection limits in the low parts per billion (ppb or ng/mL) range [24]. An imputed value was placed in the analyte results field for analytes with analytic results below the lower limit of detection; 0.10/square root of 2 = 0.07 was the lower limit of detection divided by the square root of 2. Thus, the LOD for each PFAS was 0.10 or 0.07. Metals Quantification Inductively coupled mass spectrometry (ICP-MS) measured metals in diluted whole blood. ICP-MS is a validated technique for analyzing metals in biological media. All data set metal analytes had the same detection limits. An imputed fill value was placed in the analyte results field for analytes below the lower limit of detection using the equation: lower limit of detection divided by the square root of 2 [23].The NHANES Laboratory Procedures Manual describes specimen collection and processing in detail [24]. The National Center for Environmental Health (NCEH) of the CDC's Division of Laboratory Sciences performed metal assays on whole blood samples for the NHANES 2007-2014. Blood metals were identified and quantified using the inductively coupled plasma mass spectrometry method No. ITB0001A. Determining Allostatic Load Levels This study's AL was determined using physiological evaluations of 10 health indicators or biomarkers. The biomarkers included systolic blood pressure (SBP), diastolic blood pressure (DBP), total cholesterol (TC), high-density lipoprotein (HDL) cholesterol, glycosylated hemoglobin (HbA1c), albumin (Alb), triglycerides (TG), body mass index (BMI), creatinine clearance (CLCR), and C-reactive protein (CRP). Measures of AL were determined by calculating the cutoffs for various biomarkers based on their distribution within the database. All biomarkers were transformed into quartiles based on the data distribution. The top 25% of the distribution for each marker was designated as high risk for (1) C-reactive protein (CRP), (2) triglycerides (TG), (3) total cholesterol (TC), (4) systolic blood pressure (SBP), (5) diastolic blood pressure (DBP), (6) body mass index, and (7) glycosylated hemoglobin. For the other markers where high risk is determined by lower values, the bottom 25% of the distribution was used. These markers included (1) urinary albumin (Alb), (2) creatinine clearance (CLCR), and (3) high-density lipoprotein (HDL) cholesterol. High risk for each marker was assigned a value of 1, with low risk assigned a value of 0 to obtain a total AL index out of 10. An AL value greater than 3/10 was considered elevated, as indicated by the prior work of the team and others [2,[25][26][27]. Data Analysis We used BKMR with the hierarchical variable selection method due to highly correlated variables and collinearity in the datasets. We utilized the BKMR model in the R program using the R package (bkmr) to simulate the dataset. In this study, the model evaluated the impacts of mixtures or multipollutant exposures (e.g., PFAS and metals such as cadmium, cobalt, cesium, molybdenum, lead, etc.) by comparing the implementation of statistical and characteristics methods using the (kmbayes) function. BKMR Modeling for Binary Outcomes Combining data sources from various samples, including probability and nonprobability samples, is appropriate when using Bayesian inference. The use of Bayesian inference has various benefits. It first enables the estimation of complicated models and the quantification of uncertainty measurements. The likelihood function can be used to analyze sample units based on probability. As the probability sample size grows, it is primarily set up to give these units priority in the posterior calculations. Third, it enables posterior estimates to be more effective and efficient than estimates obtained from tiny probability-only samples, with less uncertainty [19,28]. We implemented kernel machine regression (KMR) for binary outcomes, as follows: The outcome variable in this study was AL. AL index values ≥ 3 were considered high risk, with values < 3 considered low risk. Those who were high risk were assigned a 1 in the dataset, while those who were low risk were assigned a 0. Binary outcomes were performed by applying the BKMR package using the probit model for convenience of computation and to overcome some of the issues that may arise in the dataset, such as collinearity under Bayesian inference [29]. Posterior inclusion probabilities (PIPs), which offer a gauge of the variable importance of each exposure, were extracted and plotted. All models were adjusted for sex, age, smoking, physical activity, ethnicity, occupation, income, alcohol consumption, education, birthplace, and time in the U.S. The analysis within this study was conducted using R software, version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria). A flow chart containing all the steps performed in the analysis can be found in Figure 1 below. Table 1 below provides the posterior inclusion probabilities (PIPs), which measure the percentage of the data that backs the inclusion of exposure or variable in the model. In other words, it quantifies the variable significance to be included within the model. The exposures to be included in the model were PFNA, PFUA, PFOA, PFHS, mercury, cesium, and molybdenum. Figure 2 shows the association between the response variable and each individual exposure included in the model, which is known as the univariate relationship. The other exposures were fixed at their median values, and the covariates were fixed as constant. This figure shows that the association of some variables is not significant or has no association with the outcome. In other words, Figure 2 below shows the univariate independentresponse association (each individual independent and dependent-AL association) by fixing the remaining exposures to their median, with the covariates being constant. The associations in Figure 2 present the relationship of exposures with responses when the model is adjusted for covariates (sex, age, smoking, physical activity, ethnicity, occupation, income, alcohol consumption, education, birthplace, and time in the U.S.). For instance, exposure to PFNA, PFUA, PFOA, PFHS, mercury, cesium, thallium, tungsten, and uranium are associated with AL, with some of these contaminants having sharper inclines, indicating different levels of exposure. Uphill on the graphs represents a higher level of exposure, and downhill shows lower levels of exposure. In other words, concentration values increase and decrease depending on the amount of exposure. Results In Table 2, the PIPs with the highest values are explored using critical sociodemographic and behavioral variables. The six highest PIPs were molybdenum, cesium, mercury, PFNA, PFOA, and PFHS. Table 3 explores mean AL levels by ethnicity and age group. This was performed to give context to the results. The results indicated that both ethnicity and age are significantly related to AL. Table 4 explores the correlation between all the critical environmental exposures in this study. The results demonstrate that the strongest correlation exists between cesium and mercury. Discussion The main PFAS have extensive half-lives in humans and are physiologically and biologically persistent. The gap in the body of knowledge on the impact of environmental pollutants on stress and health is partly filled by attempting to understand the relationship between the cumulative physiological burden of stress (AL) and PFAS and metals [30]. This is especially true because stressors are constantly present in people's lives, and the cumulative effect on health is apparent when resilience is lacking [30,31]. BKMR provides a way to address the potential multicollinearity among numerous PFAS and metal exposures, which cannot be resolved using traditional regression modeling. Based on a comprehensive analysis of the NHANES 2007-2014 data, we assessed the relationships between metal and PFAS exposures and AL among a nationally representative sample of adults. The study's findings supported the main hypothesis, which stated that exposure to a combination of PFAS and metals is strongly linked to AL. This expands prior work by the team, which found that metals and PFAS are associated with AL using simpler modeling techniques [2,30]. In this study, combined exposure analyses of PFAS and metals showed a significant positive association between mixed PFAS and metal exposure and AL, to which cesium, molybdenum, PFHS, PFNA, and mercury contributed the most (PIP = 1, 1, 0.854, 0.824, and 0.807, respectively). In addition, the correlation between selected metals and PFAS (Table 4), with some negatively and others positively associated, suggests that the relationships between these factors are varied and require dynamic modeling techniques to capture the combined relationship appropriately. In the BKMR model, a substantial positive association between combined metal and PFAS exposure and AL existed for PFNA, PFUA, PFHS, thallium, and tungsten. The univariate relationship between AL and each exposure in the model is depicted in Figure 2. All other exposures and covariates were held constant at their respective median values. The results demonstrated which variables, in combination, were not significantly associated with AL. These models were adjusted for confounding factors, and the associations between exposures and responses became clear. Exposures to PFNA, PFUA, PFOA, PFHS, mercury, cesium, thallium, tungsten, and uranium, to name a few, are all associated with AL; some of the graphs had steeper slopes than others, reflecting the fact that there were varying degrees of association between variables. The molecular processes or toxicological pathways that underlie the relationships between human exposure to PFAS and metals and AL are not fully understood. The means by which exposure to PFAS and metals brings forth adverse health outcomes may be via AL. Simply put, AL may be the mediator between exposure to multiple contaminants and adverse health outcomes [2,32], such as heart disease, high blood pressure, metabolic syndrome, obesity, and arthritis [33]. Table 2 shows that the mean levels of the contaminants of interest varied by ethnicity; for example, Asians had high mean levels of molybdenum, cesium, and PFUA, with values of 65.9, 537, and 0.26, respectively. Blacks had higher mean mercury levels, and Whites had higher PFOA and PFHS levels than the other groups. These varied exposure levels by ethnicity speak to the variability of the contaminants of interest and the dynamism of exposure in various environments. Within our results, compared to those of the White, Asian, and Hispanic ethnicities, non-Hispanic Blacks had greater rates of high AL. Tables 2 and 3 demonstrate that across all age groups, high stressors in addition to lower levels of resilient behavior, such as physical activity, exist. This may play a role in adverse health outcomes driven by AL. Understanding the social implications of AL may help explain some of the results of this study. For instance, many ethnic groups in the U.S. experience prejudice, face poor wage employment disproportionately, and are susceptible to chronic stress [34]. In the context of multiple environmental exposures, these factors may play a role in promoting AL. When this is intertwined with inadequate healthcare, the health burden on communities exposed to combinations of exposures and health outcomes is vast [34]. Non-Hispanic Whites in the US often have lower levels of AL than minority ethnic groups, as demonstrated in Table 3, across all age groups [35]. This may partly explain the lower disease burden within this group compared to the other groups. Age is a critical variable in AL levels, with younger people typically having lower AL levels than older persons [36]. Our results, as shown in Table 3, confirm this. Continuous stressor exposure over the course of a lifetime can promote inflammation and oxidative stress, which can lead to physiological impairment and promote disease [37]. Among these is cardiovascular disease, the leading killer in the U.S. and in the world [38]. As people become older, their biological sensitivities to chronic stress vary, and the body's physiological response system also changes naturally. As a result, biological regulation may deteriorate over time, which may result in an unhealthy physical state. This scenario has the potential to cause mortality over time, especially in elderly people [35]. The literature on AL by sex varies. Some research has shown that AL levels are often lower in men who hold professional positions, such as managers and directors. On the contrary, Rogers et al. reported that men with higher levels of education are likely to have higher AL [39]. According to several studies, women who obtained higher levels of education and simultaneously held professional jobs as managers had a higher prevalence of AL [40]. People who experience continual stress due to issues such as unemployment and poverty are more likely to engage in excessive drinking, smoking, and eating, which leads to obesity, poor sleep, and, of course, increased AL [41]; our results in Tables 2 and 3 support this. According to a study by Petrovic et al., drinking, smoking, and eating too much sodium were all associated with a higher risk of developing AL. Meanwhile, physical activity and a vegetarian diet were linked to a reduction in AL [40]; our results in Table 2 support these findings. Very few laboratory studies have examined combined exposure to PFAS and metals. Therefore, future experimental and human investigations are required to further corroborate our findings and to investigate the probable mechanism for the health impacts of PFAS and metal exposure on AL, given the dearth of laboratory data and the cross-sectional design of our study. The limitation of this design means that temporality cannot be inferred. A longitudinal study would offer better insight into these exposures and health outcomes. Conclusions PFAS and other toxicants, such as metals, interact in the human body to produce AL. The mixture of PFAS and metals is critical to understand, as they may, in combination, bring forth adverse health outcomes via AL. When PFAS are found in the body alongside metals, our results indicate that their combined toxicity needs to be considered, with cesium, molybdenum, mercury, PFHS, and PFNA especially being of concern. More research is required into this matter. Research into the levels of exposure to multiple pollutants required to bring about AL must be explored if we are to gain an understanding of the realworld mixture concentrations that bring forth disease. This is of paramount significance for at-risk communities because their members lack the resources to effectively manage stress and/or avoid exposure to environmental contaminants. Funding: This research was funded by NHLBI grant R25 HL105400 and the BCSP Foundation. Institutional Review Board Statement: This study did not require IRB approval because de-identified secondary data were used. In the collection of the data by the Centers for Disease Control and Prevention, the study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of the Centers for Disease Control and Prevention. Informed Consent Statement: Not applicable. Data Availability Statement: The NHANES dataset is publicly available online, accessible at cdc. gov/nchs/nhanes/index.htm (accessed on 12 January 2023). Conflicts of Interest: The authors declare no conflict of interest.	2023-03-24T15:23:06.653Z	2023-03-01T00:00:00.000	{ "year": 2023, "sha1": "ea8c26d0b0f42cb0a82e906c74e5cdc1db2463d4", "oa_license": "CCBY", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "7a8620d1a9a40e543b105f2f29b84eb2b55b3c70", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [] }
203565687	pes2o/s2orc	v3-fos-license	The Importance of microRNAs in RAS Oncogenic Activation in Human Cancer microRNAs (miRNAs) regulate gene expression by modulating the translation of protein-coding RNAs. Their aberrant expression is involved in various human diseases, including cancer. Here, we summarize the experimental pieces of evidence that proved how dysregulated miRNA expression can lead to RAS (HRAS, KRAS, or NRAS) activation irrespective of their oncogenic mutations. These findings revealed relevant pathogenic mechanisms as well as mechanisms of resistance to target therapies. Based on this knowledge, potential approaches for the control of RAS oncogenic activation can be envisioned. INTRODUCTION microRNAs (miRNAs) are small (19-24 nucleotides) non-coding RNAs discovered in 1993 in studies related to embryonic development of C. elegans (1,2). Their importance significantly increased following the discovery of their existence in all eukaryotic organisms (3). Currently, 2,654 mature miRNAs, originating from 1917 precursors, are described in humans (http://www.mirbase. org/) (4,5). Their main function is to negatively regulate gene expression at the post-transcriptional level through the interaction of their "seed" portion by sequence homology typically with the 3 ′ non-coding regions of messenger RNAs (mRNAs). Through this interaction, miRNAs limit translation, or promote degradation of target mRNAs (6,7). The modulation of target mRNAs by miRNAs is complex, considering that each mRNA is generally targeted by multiple miRNAs, and the strength of this interaction is variable (8). Classically, it has been thought that each miRNA can interact with hundreds of target mRNAs. However, recent reports have highlighted RNA transcripts inducing degradation of respective interacting miRNAs through a mechanism known as "target-directed miRNA degradation" (TDMD) (9,10). Added to the complexity of these direct interactions is the fact that some long non-coding RNA (lncRNA) could function as "sponges, " that act as a buffer and prevent the action of miRNAs on target protein-coding mRNAs (11,12). Lastly, it is also important to consider that cell co-localization of each miRNA with the target mRNAs is necessary and depends on the eventual tissue-specific expression of each of the interacting RNAs. Thus, miRNAs, taken together, represent an essential phase in the regulation of gene expression by modulating the translation of the entire transcriptome (13,14). Given their biological importance, their deregulation plays a significant role in pathogenic mechanisms, including the neoplastic transformation (15,16). The first evidence associating miRNAs with human malignant diseases was the discovery of miR-15 and miR-16 in the minimal region of deletion at chromosome 13q14 in chronic lymphatic leukemia (17). Since this seminal study, a myriad of other studies has confirmed the role of miRNAs in tumorigenesis and other human diseases as well. miRNAs AS DIRECT REGULATORS OF RAS The first functional evidence to establish a molecular link between the deregulation of miRNAs with an explicit oncogenic pathway was published in 2005 when Slack and collaborators reported the importance of the downregulation of members of the let-7 miRNA family with the activation of oncogenes of the RAS family (18). The study demonstrated that the 3 ′ UTRs of KRAS, NRAS and HRAS mRNAs comprised multiple complementary let-7a binding sites. The enforced expression of let-7 could indeed reduce RAS protein levels (18). Conversely, let-7 downregulation could lead to the loss of its post-transcriptional control, causing RAS over-expression and activation. This study was decisive in proving that aberrant expression of miRNAs could play an important role in tumor initiation and progression, and paved the way for studies that extended miRNA involvement to all phases of neoplastic initiation and progression (19). The involvement of RAS (KRAS, NRAS, HRAS) in human tumors is mainly associated with the presence of activating mutations at codons 12, 13 and 61, able to activate various molecular pathways, which play a key role in a large number of tumor traits, spanning from cell proliferation, cell survival, cytoskeleton organization, motility, and more (20). The demonstration of the role of miRNAs in the abnormal regulation of RAS thus represented another important mechanism involved in key steps of tumorigenesis. Since then, quite a few other reports have demonstrated the modulation of RAS by miRNAs. In many cases, the interaction was only predicted by computer algorithms, but several studies have experimentally validated these interactions. Table 1 lists the microRNAs for which the ability to modulate the expression of KRAS, NRAS, or HRAS has been experimentally confirmed. As mentioned, let-7 was the first, and probably the most important miRNA implicated in the regulation of genes of the RAS family (18). In the human genome, 12 loci are known to encode for members of the let-7 family: let-7a-1, -2, -3; let-7b; let-7c; let-7d; let-7e; let-7f-1, -2; let-7g; let-7i; miR-98. While it is described that members of the let-7 family are upregulated in the course of cell differentiation, numerous studies have reported the reduction of let-7 expression in different tumor types (21,22). Already in 2004, Takamizawa et al. demonstrated the downregulation of let-7 in non-small cell lung carcinoma (NSCLC) (23,24) and documented its prognostic significance. Furthermore, in line with these observations, they proved that enforced expression of let-7 miRNA could inhibit in vitro cell growth of the lung adenocarcinoma A549 cells (23,(25)(26)(27). These studies were further confirmed in murine in vivo models of NSCLC (28,29) and revealed that let-7 mimics could represent potential therapeutic molecules. (HMGA), Janus protein tyrosine kinase (JAK), signal transducer and activator of transcription 3 (STAT3) (30). Its action as a tumor suppressor gene is therefore achieved through the ability to interact with multiple oncogenes and inhibit the activation of their molecular pathways (18,28). Essentially all types of human cancer present a general downregulation of let-7 (21). Among others, the modulation of RAS by let-7 was demonstrated in colorectal cancer (CRC) where let-7 is strongly down-regulated in tumor tissues compared to adjacent healthy tissues. Similar to the study on NSCLC cells, let-7 was also shown to act as a growth suppressor in human CRC cells (31). Confirming the importance of RAS regulation by let-7, the discovery of the LCS6 polymorphism (Let-7 Complementary Sites 6, rs61764370) in the KRAS 3 ′ UTR region further demonstrated let-7 expression altering interaction. This polymorphism has been associated with a greater risk of developing tumors and worse prognosis in lung, oral, and colorectal cancer (32)(33)(34). An understanding of a mechanism leading to let-7 downregulation in cancer came from studies on LIN28 in mammals. Lin28 and Lin28b act as RNA binding proteins that are able to associate with the terminal loop of the precursors of let-7 family miRNAs and block their processing into mature miRNAs (35)(36)(37)(38). Since LIN28 is over-expressed in human cancer, this mechanism causes let-7 down-regulation, which establishes a connection with RAS and other cancer-associated signalings. Let-7 is not the only miRNA involved in the regulation of RAS (HRAS, KRAS, or NRAS) (39). Among the miRNAs involved in the regulation of members of the RAS family, miR-143 and miR-145, co-expressed in the same primary transcript, can target both KRAS and NRAS, and have been found to be down-regulated in numerous human tumors (40)(41)(42). Already in 2003 Michael et al. documented a significant reduction of miR-145 in CRC compared to normal mucosa (43) and in 2014, Pagliuca et al. confirmed that the miR-143/miR-145 cluster, highly expressed in normal colon, was significantly decreased in CRC (44). Their reduced expression has been correlated with p53 mutations capable of reducing the maturation process of these miRNAs (45). Very similar to let-7, members of the miR-181 family were shown to target all the RAS family members (HRAS, KRAS, and NRAS). They were found downregulated in different types of cancer, such as oral squamous cell carcinoma (46,47), gastric cancer (48), and gliomas (49). These findings suggest that miR-181 down-regulation is one of the mechanisms leading to oncogenic RAS activation in these tumors. It is notable that in spite of KRAS activation by gene mutation in 90% of the cases in pancreatic cancer, various miRNAs capable of directly targeting KRAS are simultaneously downregulated. Specifically, miR-96, miR-126, and miR-217 (50)(51)(52)(53). Since the reduced expression of these miRNAs correlates with higher KRAS expression, these alterations likely represent a mechanism for strengthening the already activated RAS signaling. Another noteworthy miRNA capable of targeting KRAS is miR-134. It was found downregulated in glioblastoma and renal cell carcinoma (54,55). miR-134 downregulation correlated with the activation of the MAPK pathway and its enforced expression in renal cancer cells could inhibit in vitro migration and invasive traits. Oncogenic mutations resulting in RAS activation are prevalent in most human tumors, but there are exceptions. RAS mutations in HCC are rare events but paradoxical wild-type RAS activation is common (56). Dietrich et al. (57) discovered that wild-type KRAS expression was increased in HCC compared to non-tumor liver and revealed an inverse correlation with miR-622 expression. In addition to the above-mentioned examples, several other miRNAs were proven to target and inhibit the expression of RAS oncoproteins ( Table 1). These miRNAs are generally downregulated in tumors, thus concurring with reciprocal overexpression and activation of RAS, irrespective of activating gene mutations. miRNAs AS RAS EFFECTORS The interplay between miRNAs and RAS is not only represented by miRNAs acting as negative modulators of RAS but also includes downstream miRNA effectors. The most significant is undoubtedly miR-21, which is up-regulated by KRAS oncogenic mutants in non-small-cell lung cancer (58), laryngeal squamous cell carcinoma (59), and pancreatic adenocarcinoma (60) as well as many other human cancers. miR-21 is a known oncomiR capable of blocking the expression of tumor suppressor genes antagonists of the PI3K-AKT pathway, such as PTEN, or of the RAS-MAPK pathway, such as PDCD4 or RASA1 (61-63) (Figure 1). miRNAs AS REGULATORS OF RECEPTOR TYROSINE KINASES (RTKS) RAS is a crucial node that connects receptor tyrosine kinases (RTKs) with downstream molecular pathways (Figure 1). Hence, miRNAs can affect RAS activity by acting on RTKs as well as MAPK, PI3K, or other pathways. It is a well-known notion that RAS activation is physiologically triggered by RTKs, a category of transmembrane receptors that become activated in response to growth factors. Several miRNAs are known to target RTK mRNAs and their dysregulation can lead to inappropriate activation of the targeted RTK. Just to mention a few examples, miR-7, miR-539 and miR-103-3p can target and modulate the expression of the epidermal growth factor receptor (EGFR) (64-66); miR-26a was shown to target c-MET, the hepatocyte growth factor receptor (67); miR-199-3p can target the vascular endothelial growth factor receptors 1 and 2 and the VEGFA ligand (68); miR-7 and miR-98 can target the insulin growth factor receptor gene (64,69). All the above-mentioned miRNAs were found dysregulated in a variety of human cancers. miR-539 is downregulated in breast cancer (BC) tissues and cell lines. miR-539 enforced expression could inhibit BC cells proliferation and tumor growth in vitro and in vivo (65). miR-7 is downregulated in breast and colorectal cancer (CRC) cells (64,66) and its reduced expression in BC patients correlated with higher stage, grade, and poor prognosis (64). The tumor suppressor activity of miR-103-3p was confirmed by the anti-proliferative effects after its enforced expression in lung cancer cell lines; furthermore, the downregulation of miR-103a-3p in NSCLC was associated with poor prognosis (66). miR-26a reduced levels were associated with poor prognosis in Hepatocellular carcinoma (HCC) (67). MiR-26a can also control the expression of VEGFA in HCC cells and impairs VEGFR2-signaling thereby controlling angiogenesis. miR-199-3p, another miRNA that can target VEGFR1, VEGFR2, and the ligand VEGFA (68), is frequently down-regulated in HCC and it has been shown to have in vitro and in vivo anti-tumor activity in HCC models (68,70). MiR-98 is down-regulated in retinoblastoma, where it also represents a prognostic marker (69). The above-reported miRNAs are just a few examples to show how their deregulation can lead to RTKs overexpression and consequently activation of RAS and its downstream pathways. The latter are themselves regulated by miRNAs, whose deregulation may directly cause the activation of RAS downstream effectors independently from RAS triggering. miRNAs AS REGULATORS OF MAPK PATHWAY The MAPK pathway is a well-studied pathway that promotes cell proliferation and is controlled by RAS activation. It includes several effectors with oncogenic function, widely studied in different types of tumors and whose mutations also represent tumorigenic mechanisms. BRAF is probably the most studied element of the MAPK pathway. BRAF activation has been associated with a missense mutation V600E, commonly found in melanoma and thyroid cancer, but also present at low frequency in several other types of human cancer (71). As expected, various miRNAs can target and regulate BRAF expression. KRAS targeting miR-143 and miR-145, that we have mentioned above, can also target BRAF, indicating a very important role of these miRNAs in regulating the MAPK pathway at several levels (44). As mentioned earlier, these miRNAs are frequently downregulated in various types of cancer. miR-9-5p is another miRNA targeting BRAF, which was shown to be down-regulated in papillary thyroid carcinoma (65). Further downstream of MAPK pathway cascade, MEK1/MEK2 (also called (MAP2K1 and MAP2K2) as well as ERK1/ERK2, are also targets of miRNAs. miR-1826 can target MEK1. It is down-regulated in bladder cancer and its reduced expression is associated with more severe pathological traits (pT and grade) (72). miR-101 can also target MEK1. This miRNA exhibits reduced expression in diffuse large B cell lymphoma (DLBCL) and it is associated with a worse prognosis (73). miR-665 has been also shown to indirectly activate MEK in BC cells by targeting the nuclear receptor subfamily 4 group A member 3 (NR4A3) gene. This miRNA is upregulated in breast cancer where its upregulation is associated with metastasis and poor survival (74). miRNAs THAT ACT ON MULTIPLE TARGETS OF THE RAS PATHWAY Among the several miRNAs that regulate elements of the RAS-centered pathways, some miRNAs target multiple genes belonging to the pathway thus reinforcing their role in modulating MAPK pathway activation. In this respect, miR-134 is a typical example, as its target genes not only include KRAS (75), but also EGFR (76), HER2 (77), STAT5B (54), and PIK3CA (78), which are upstream and downstream elements of the RAS-centered pathways. This miRNA is downregulated in numerous types of human cancers, where it affects cell proliferation, survival, invasiveness, metastasis, and apoptosis [reviewed in (79)]. This miRNA exemplifies the deregulatory action of single miRNA and consequent wide effects on tumorigenic signals by acting on multiple elements of the RAS pathways (79). Other miRNAs targeting multiple RAS effectors include miR-143 / miR-145, previously mentioned to target all RAS genes and BRAF; miR-524-5p that can target both BRAF and ERK2. In melanoma, miR-524-5p is downregulated and affects cell migration and proliferation both in vitro and in vivo (80). These miRNAs are potentially very important, as they can represent useful molecules to effectively restore the normal expression of multiple proteins belonging to RAS pathways. microRNAs IMPLICATED IN RESISTANCE TO TARGET THERAPIES Therapeutic interventions in advanced cancers include traditional chemotherapy as well as targeted/immuno-therapies. Targeted therapies make use of molecules capable of blocking aberrantly activated oncogenes that act as tumorigenic drivers. Oncogenic RAS proteins would represent outstanding targets for such therapies. But, no drug targeting RAS has been yet validated for clinical use. At present, most available targeted therapies are instead designed to block the activity of several elements of RAS-centered pathways. These include a large number of tyrosine kinase inhibitors (TKIs) or antibodies against RTKs; drugs that target BRAF V600E mutation (vemurafenib and dabrafenib), MEK (trametinib, cobimetinib and binimetinib), PI3K mutations (alpelisib), and mTOR (everolimus). The RAS pathways are therefore targeted by several drugs, with the RAS itself being a major exception. Even more disappointing is the fact that mutant activated RAS often reduces the efficacy of targeted drugs and patients become resistant to therapies. One of the best-known mechanisms associated with the emergence of TKI resistance is indeed KRAS mutation. It is known that tumors with KRAS mutations at codons 12, 13, 61, or 146 do not respond to treatment with anti-EGFR antibodies or TKIs and therefore mutational analyses on all RAS genes are carried out on tumor biopsies before a therapeutic regimen is chosen. Among miRNAs that target KRAS, the reduced expression of miR-181a was shown to be associated with gefitinib resistance in lung cancer (96,97); in CRC patients treated with cetuximab, it was reported that low levels of miR-181a were associated with a lower overall survival, indicating a reduced efficacy of anti-EGFR therapy (98). While miR-145 was shown to synergize with cetuximab activity (99), high levels of let-7 could predict the efficacy of cetuximab therapy even in CRC patients carrying mutant KRAS (100). Dietrich et al. (57) not only revealed an inverse correlation of KRAS and miR-622 expression but, additionally, they could attribute KRAS-miR-622 interplay to therapy resistance since sorafenib induced further KRAS augmentation and downregulation of miR-622. These few examples suggest that the miRNA-mediated modulation of RAS protein levels can indeed affect the response to TKIs or anti-EGFR targeted therapies. In addition to RAS, the dysregulation of miRNAs responsible for the activation of elements of the MAPK or the PI3K pathways can also reduce the efficacy of TKIs. For example, the reduction of PTEN protein level by up-regulated miRNAs, like miR-21, miR-221, miR-23a and miR-214, can reduce efficacy of TKIs in lung cancer by activating the PI3K pathway (83,(101)(102)(103)(104)(105)(106). miRNAs have also been associated with resistance to the BRAF inhibitor vemurafenib (107)(108)(109)(110). In short, several studies have shown that the dysregulation of miRNAs has an important role in the efficacy of target therapies, thus suggesting that their levels of expression can be useful to guide the choice of therapy, alongside the more conventional mutational investigations. Furthermore, they also provide suggestions for potential therapeutic approaches useful to restore or improve sensitivity to treatments. CONCLUSIONS Taken together, published data provides a strong indication that altered miRNA expression represents an important mechanism for RAS activation, with various implications. First, it represents a mechanism of pathogenic relevance, responsible for the promotion of several tumor traits, irrespective of RAS oncogenic mutations. Second, considering that the activation of RAS represents a frequent mechanism of resistance for drugs directed against RTKs, it is possible that miRNA dysregulation represents a relevant aspect to consider when assessing the proper management of patients on target therapies. Third, miRNAs may represent potentially useful molecules for the control of RAS oncogenic activation, aimed at overcoming the lack of drugs targeting RAS and possibly improving the efficacy of target therapies.	2019-09-28T13:06:02.314Z	2019-09-27T00:00:00.000	{ "year": 2019, "sha1": "bf523152b0416e30c98441632f265cc9d3e8ccc4", "oa_license": "CCBY", "oa_url": "https://www.frontiersin.org/articles/10.3389/fonc.2019.00988/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "bf523152b0416e30c98441632f265cc9d3e8ccc4", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
244330006	pes2o/s2orc	v3-fos-license	K-means Cluster Algorithm Applied for Geometric Shaping Based on Iterative Polar Modulation in Inter-Data Centers Optical Interconnection : The demand of delivering various services is driving inter-data centers optical interconnection towards 400 G/800 G, which calls for increasing capacity and spectrum efﬁciency. The aim of this study is to effectively increase capacity while also improving nonlinear noise anti-interference. Hence, this paper presents a state-of-the-art scheme that applies the K-means cluster algorithm in geometric shaping based on iterative polar modulation (IPM). A coherent optical communication simulation system was established to demonstrate the performance of our proposal. The investigation reveals that the gap between IPM and Shannon limit has signiﬁcantly narrowed in terms of mutual information. Moreover, when compared with IPM and QAM using the blind phase searching under the same order at HD-FEC threshold, the IPM-16 using the K-means algorithm achieves 0.9 dB and 1.7 dB gain; the IPM-64 achieves 0.3 dB and 1.1 dB gain, and the IPM-256 achieves 0.4 dB and 0.8 dB gain. The robustness of nonlinear noise and high capacity enable this state-of-the-art scheme to be used as an optional modulation format not only for inter-data centers optical interconnection but also for any high speed, long distance optical ﬁber communication system. Introduction With the massive growth of cloud computing, 5G/6G, web-based applications, and other new types of services in recent years, the Internet traffic has been steadily expanding, which promotes optical interconnection networks (OINs) developing towards 400 G/800 G [1,2]. In this trend, an optical transmission system with high spectral efficiency is required to accommodate a higher transmission rate. There are numerous strategies to significantly increase transmission rate, among which advanced coding and modulation, multi-dimensional multiplexing and forward error correction (FEC) coding play a vital role [3,4]. Researchers have conducted substantial research on the topic of increasing system capacity. In terms of multi-dimensional multiplexing, ref. [5] proposed a method to realize a space-division multiplexing network in data center to overcome the optical network capacity crunch by using multi-core optical fiber. It performs effectively in data centers with short distances. However, it is unfavorable for OINs since inter-core crosstalk is becoming increasingly severe in long-haul transmission. Another significant aspect of improving capacity is using advanced coding and modulation. The author of [6] proposed a novel 64, 256, respectively, the BER performance of the proposed scheme is compared with QAM adopting blind phase searching (BPS) in the same order. The results reveal that the proposed scheme has a lower BER, suggesting that our proposal can enhance capacity without sacrificing BER performance. The paper is organized as follows. We begin in Section 2 by discussing the principles of IPM and K-means and generate the IPM constellations of different orders. In Section 3, we describe the configurations of the MATLAB simulation system and then evaluate the simulation results. We discussed the potential application scope in Section 4. Finally, we provide some concluding remarks in Section 5. Geometric Shaping Based on Iterative Polar Modulation Geometric shaping could achieve shaping gain by optimizing the shape of the highdimensional signal constellation. Most of the GS algorithms are based on optimizing a specific objective function, such as maximizing mutual information, minimizing BER, or minimizing mean square error (MMSE) to determine the location of constellation points satisfying some conditions. The IPM modulation format considers the MMSE based on iterative polar quantization procedure [17]. The constellation coordinates determined by a numerical algorithm like Arimoto-Blahut [22] are optimum in the transmission system with significant influence of thermal noise and amplified spontaneous emission noise. The nonuniform iterative polar quantization considers implementing MMSE in two dimensions, including scalar nonuniform amplitude (r) and scaler uniform phase quantization (ϕ). The calculation formula of MSE in polar quantization consists of two parts, and could be computed as Equation (1). where D granul represents the granulation noise and D overload represents the overload noise. The granulation noise and overload noise are given by Equations (2) and (3). where L i is the number of constellation points on i-th circle; L r is the number of circles; θ j,i is the phase of j-th points on i-th circle; p r (r) represents the source of probability density function (PDF). m i is the optimum radius of i-th circle. The optimum number of constellation points of i-th circle is determined by using Lagrange multiplier method to implement MMSE. Accordingly, we have As for PDF function of the optimum source, we consider the case of Gaussian distribution, that is We assumed the partial derivatives to m i and r, respectively, of Equations (2) and (3) are equal to zero, then the optimum radius is given by Repeat calculating Equations (4) and (5) until these equations satisfied the following limits of integration determined by Equation (7). We could obtain the optimum IPM constellation coordinates using the above Equations (2)- (7). In this paper, three kinds of constellations are generated based on nonuniform iterative polar quantization; the orders of these constellations M are equal to 16, 64, 256, respectively. The relevant constellation is shown in Figure 1, which is consistent with [17][18][19][20]. The IPM constellations show that it is a kind of circle modulation method. Still, the performance is better than circle QAM (CQAM) because of its MMSE and larger channel capacity, especially in the IPM-256. We find out that there is a point at the origin of the coordinate axis in constellations of IPM-16 and IPM-256, respectively, as shown in Figure 1a,c, respectively, which makes it dramatically reduce the average transmission power, which could be called centered-IPM (CIPM). We assumed the partial derivatives to and , respectively, of Equations (2) and (3) are equal to zero, then the optimum radius is given by Repeat calculating Equations (4) and (5) until these equations satisfied the following limits of integration determined by Equation (7). We could obtain the optimum IPM constellation coordinates using the above Equations (2)- (7). In this paper, three kinds of constellations are generated based on nonuniform iterative polar quantization; the orders of these constellations are equal to 16, 64, 256, respectively. The relevant constellation is shown in Figure.1, which is consistent with [17][18][19][20]. The IPM constellations show that it is a kind of circle modulation method. Still, the performance is better than circle QAM (CQAM) because of its MMSE and larger channel capacity, especially in the IPM-256. We find out that there is a point at the origin of the coordinate axis in constellations of IPM-16 and IPM-256, respectively, as shown in Figure 1a and Figure 1c, respectively, which makes it dramatically reduce the average transmission power, which could be called centered-IPM (CIPM). K-means Cluster Algorithm at the Receiving end of Data Center IPM signals are impacted by ASE, fiber dispersion, and other phase noise distortions while they are transmitted from one DC to another. The nonlinear phase noise caused by the Kerr effect will dominate the modulated signal when OSNR is high sufficient. Kerr effect is an electro-optic effect, which indicated that the refractive index is proportional to the square of the applied electric field. Since the refractive index is nonlinear, as the electric field intensity varies in the optical fiber, the refractive index fluctuates, and the signal phase shifts as well, resulting in nonlinear phase noise [23][24][25]. To simulate the nonlinear phase noise generated by the Kerr effect during optical pulse propagation, the Generalized Nonlinear Schrodinger Equation (GNLSE) can be used as the mathematical modeling to describe the process of optical pulse propagation equation [26]. GNLSE, on the other hand, cannot provide an analytical solution for any input light pulse. As a consequence, a numerical method called Split-Step Fourier Transform (SSFT) should be employed to model the propagation of light pulses in a single-mode fiber (SMF) [27]. Accordingly, the essential DSP module is required to recover optical signals at the receiving end of the data center. In response to the frequency offset generated by laser and phase rotation caused by nonlinear phase noise (NLPN), the most common use recovery algorithm is BPS. However, the complexity of BPS grew dramatically as the modulation K-means Cluster Algorithm at the Receiving End of Data Center IPM signals are impacted by ASE, fiber dispersion, and other phase noise distortions while they are transmitted from one DC to another. The nonlinear phase noise caused by the Kerr effect will dominate the modulated signal when OSNR is high sufficient. Kerr effect is an electro-optic effect, which indicated that the refractive index is proportional to the square of the applied electric field. Since the refractive index is nonlinear, as the electric field intensity varies in the optical fiber, the refractive index fluctuates, and the signal phase shifts as well, resulting in nonlinear phase noise [23][24][25]. To simulate the nonlinear phase noise generated by the Kerr effect during optical pulse propagation, the Generalized Nonlinear Schrodinger Equation (GNLSE) can be used as the mathematical modeling to describe the process of optical pulse propagation equation [26]. GNLSE, on the other hand, cannot provide an analytical solution for any input light pulse. As a consequence, a numerical method called Split-Step Fourier Transform (SSFT) should be employed to model the propagation of light pulses in a single-mode fiber (SMF) [27]. Accordingly, the essential DSP module is required to recover optical signals at the receiving end of the data center. In response to the frequency offset generated by laser and phase rotation caused by nonlinear phase noise (NLPN), the most common use recovery algorithm is BPS. However, Electronics 2021, 10, 2417 5 of 12 the complexity of BPS grew dramatically as the modulation order increased. Additionally, the BPS algorithm could only compensate (−π/4, π/4) phase shift of the range, which is clearly insufficient when there is a factor of frequency offset. K-means cluster algorithm is an unsupervised learning, which will classify comparable objects into the same cluster. It has the advantages of low complexity, easy implementation, and fast convergence. Its good classification characteristics could be employed to demodulate optical signals that have been greatly affected by NLPN and avoid using BPS algorithm with high complexity. The principle of K-means is mainly to figure out the associated centroid according to the sum of minimum Euclidean distance. The specific procedure of K-means is as follows. Firstly, randomly choose K points as the initial centroids; we set the initial centroids here as the coordinates of the transmission constellation to make the algorithm rapidly converge to the optimal. Then calculate the Euclidean distance between each point and K initial centroids, and assign which clusters the point belongs to. After resetting the locations of the centroid, repeat the above steps until the centroid will not change. As an example, the transmitting and receiving procedure for 64-QAM is as follows. The bit stream to be transmitted is modulated into optical 64-QAM signal at the transmitter, and then the numerical solution of GNLSE using SSFT algorithm to simulate the optical pulse propagation in SMF is calculated. N symbols affected by dispersion and NLPN are received at the receiving end and classified using the principle of the above-mentioned K-means. The acquired centroid is one-to-one corresponding to the points of the ideal 64-QAM constellation to complete the demodulation. Figure 2 shows the cluster classification result after deploying the K-means algorithm to 64-QAM; it is evident that even if the optical signal phase rotated seriously, the algorithm could still classify and demodulate constellation points more accurately. Electronics 2021, 10, x FOR PEER REVIEW 5 of 1 order increased. Additionally, the BPS algorithm could only compensate (− 4 ⁄ , 4 ⁄ phase shift of the range, which is clearly insufficient when there is a factor of frequency offset. K-means cluster algorithm is an unsupervised learning, which will classify comparable objects into the same cluster. It has the advantages of low complexity, easy implementation, and fast convergence. Its good classification characteristics could b employed to demodulate optical signals that have been greatly affected by NLPN and avoid using BPS algorithm with high complexity. The principle of K-means is mainly to figure out the associated centroid according to the sum of minimum Euclidean distance The specific procedure of K-means is as follows. Firstly, randomly choose K points as th initial centroids; we set the initial centroids here as the coordinates of the transmission constellation to make the algorithm rapidly converge to the optimal. Then calculate th Euclidean distance between each point and K initial centroids, and assign which cluster the point belongs to. After resetting the locations of the centroid, repeat the above step until the centroid will not change. As an example, the transmitting and receiving procedure for 64-QAM is as follows The bit stream to be transmitted is modulated into optical 64-QAM signal at th transmitter, and then the numerical solution of GNLSE using SSFT algorithm to simulat the optical pulse propagation in SMF is calculated. N symbols affected by dispersion and NLPN are received at the receiving end and classified using the principle of the above mentioned K-means. The acquired centroid is one-to-one corresponding to the points o the ideal 64-QAM constellation to complete the demodulation. Figure 2 shows the cluste classification result after deploying the K-means algorithm to 64-QAM; it is evident tha even if the optical signal phase rotated seriously, the algorithm could still classify and demodulate constellation points more accurately. The Establishment of the Simulation System To verify the performance of the K-means cluster algorithm applied for the IPM modulation format scheme, the simulation system of a single carrier system for the dat center interconnection network was established, as shown in Figure 3. Additionally, th parameters of the system are listed in Table 1. At the transmitter, the iterative pola quantization procedure was run to generate the transmitted initially constellations. W generate three constellations here, and the order of these constellations is equal to 16 The Establishment of the Simulation System To verify the performance of the K-means cluster algorithm applied for the IPM modulation format scheme, the simulation system of a single carrier system for the data center interconnection network was established, as shown in Figure 3. Additionally, the parameters of the system are listed in Table 1. At the transmitter, the iterative polar quantization procedure was run to generate the transmitted initially constellations. We Electronics 2021, 10, 2417 6 of 12 generate three constellations here, and the order of these constellations M is equal to 16, 64, 256, respectively. Random pseudo-random binary sequences (PRBS) are generated in MATLAB and sent to the modulator to realize IPM modulation. The modulated signals are shaped by root raised cosine filter, which roll-off factor is equal to 0.25. After pulse shaping, signals are split up into two streams, amplified by electrical amplifiers (EA), and sent to one of the branches of Mach-Zehnder Modulator (MZM). Another branch is driven by External Cavity Laser (ECL), which operates at 1550 nm and its linewidth is 0.1 MHz. Polarization beam splitter (PBS) divide the optical source into two orthogonal polarization beams. Those two orthogonal polarization beams and two streams of data signals interact on MZMs to produce the modulated optical signals combined as optical wave by polarization beam combiner (PBC). Multi-span fiber link consists of recirculation loop, single-mode (SMF) fiber with a length of 100 km, and erbium-doped fiber amplifier (EDFA) with the gain of 20 dB. The main parameters of SMF are chromatic dispersion (CD), attenuation, and nonlinear coefficient, which is equal to 20 ps/(nm·km), 0.2 dB/km, and 1.3 (W·km) −1 , respectively. At the receiving end, an optical source generated by another ECL drives the balanced photodetector (PD) after 90 • hybrid to complete the process of coherent receiving. The primary affection influence by SMF in a single carrier system is chromatic dispersion and self-phase modulation. Therefore, we need the offline DSP module to recover signals and calculate the BER. The process of DSP is shown in the block diagram at the bottom of Figure 3, which consists of normalization using Gram-Schmidt orthogonalizing process (GSOP), clock recovery using Gardner algorithm, CD compensation, adaptive equalization using constant modulus algorithm (CMA), and phase recovery using BPS or K-means. The recovery signals are demodulated, and then the MI and BER of the transmission are acquired. Results and Discussion To study the performance of channel capacity of different orders of IPM and QAM modulation format, we quantify it by using the physical amount of mutual information. MI versus SNR of different modulation formats after 100 km transmission is shown in Figure 4a. The line with black diamonds is the Shannon capacity. We can find out that MI in IPM outperforms QAM modulation. Moreover, the larger the modulation order is, the closer the distance IPM is to the Shannon Capacity, especially in IPM-256, the purple line in Figure 4a. The MI versus recirculation loops are shown in Figure 4b. We consider the range of the loop is 1 to 10, length of each loop is 100 km. The results show that when employing IPM-16 and QAM 16, the MI is almost the same at loop = 1 and 2. When transmission distance is small, the nonlinear effect is not accumulated too much; when increasing the loop, the MI of IPM-16 is larger than QAM-16, which indicates that IPM-16 could carry more information. When M = 64, the capacity of IPM outperforms QAM even at the larger loop. It indicates that the average power of IPM-64 has a much lower value and has good tolerance to the dispersion, nonlinear effect, and other noises of SMF. When M = 256, the MI of IPM outperforms QAM at loop = 1 to 4; the performance of IPM is not better than QAM when loop > 4. However, as we all know, the distance between DC is usually no more than 100 km. Therefore, the IPM-256 is also suitable for the data center interconnection network. Order of constellation 16, 64, 256 The primary affection influence by SMF in a single carrier system is chromatic dispersion and self-phase modulation. Therefore, we need the offline DSP module to recover signals and calculate the BER. The process of DSP is shown in the block diagram at the bottom of Figure 3, which consists of normalization using Gram-Schmidt orthogonalizing process (GSOP), clock recovery using Gardner algorithm, CD compensation, adaptive equalization using constant modulus algorithm (CMA), and phase recovery using BPS or K-means. The recovery signals are demodulated, and then the MI and BER of the transmission are acquired. Results and Discussion To study the performance of channel capacity of different orders of IPM and QAM modulation format, we quantify it by using the physical amount of mutual information. MI versus SNR of different modulation formats after 100 km transmission is shown in Figure 4a. The line with black diamonds is the Shannon capacity. We can find out that MI in IPM outperforms QAM modulation. Moreover, the larger the modulation order is, the closer the distance IPM is to the Shannon Capacity, especially in IPM-256, the purple line in Figure 4a. The MI versus recirculation loops are shown in Figure 4b. We consider the range of the loop is 1 to 10, length of each loop is 100 km. The results show that when employing IPM-16 and QAM 16, the MI is almost the same at loop = 1 and 2. When transmission distance is small, the nonlinear effect is not accumulated too much; when increasing the loop, the MI of IPM-16 is larger than QAM-16, which indicates that IPM-16 could carry more information. When M = 64, the capacity of IPM outperforms QAM even at the larger loop. It indicates that the average power of IPM-64 has a much lower value and has good tolerance to the dispersion, nonlinear effect, and other noises of SMF. When M = 256, the MI of IPM outperforms QAM at loop = 1 to 4; the performance of IPM is not better than QAM when loop > 4. However, as we all know, the distance between DC is usually no more than 100 km. Therefore, the IPM-256 is also suitable for the data center interconnection network. The fiber channel model can be represented as a linear dispersion noise, additive white Gaussian noise (AWGN) and nonlinear phase noise channel [28,29]. The iterative polar quantization process considers AWGN, such as thermal and ASE noise, rendering the position distribution of constellation points is more subject to the channel characteristics. Hence, the MI can reach its maximum value and the geometric shaping The fiber channel model can be represented as a linear dispersion noise, additive white Gaussian noise (AWGN) and nonlinear phase noise channel [28,29]. The iterative polar quantization process considers AWGN, such as thermal and ASE noise, rendering the position distribution of constellation points is more subject to the channel characteristics. Hence, the MI can reach its maximum value and the geometric shaping gain is realized. The aforementioned are the major reasons which explain how the suggested technique can effectively narrow the gap with the Shannon limit. We evaluate the BER performance of IPM-16 and QAM-16 with K-means or BPS in different SNR ranges at loop = 5, as shown in Figure 5a, and the recovery constellations SNR = 16 dB are shown in Figure 5b. The BER decreases as the SNR increases, and it is evident that IPM-16 with K-means algorithm outperforms IPM without K-means and QAM-16. The hard-decision FEC (HD-FEC) threshold is 3.8 × 10 −3 , which is mainly used to evaluate the BER performance. Specifically, when SNR is small, the performance of these three schemes is approximately equal. Even the performance of QAM-16 is better than IPM at SNR = 10 to 13. With the increase in SNR, the advantages of the IPM scheme gradually appear. It is clear that IPM with K-means reach the HD-FEC threshold at SNR = 16, while the SNR of reaching the threshold of IPM without K-means and QAM-16 is 16.9 dB, 17.7 dB, respectively. Therefore, the gain of proposed schemes is 0.9 dB, 1.7 dB compared with IPM-16 without K-means cluster algorithm and QAM-16. In the diagram of recovery constellations, the black cross is the centroid locations after a serious iteration, and the centroid could provide a reference for subsequent demodulation. gain is realized. The aforementioned are the major reasons which explain how the suggested technique can effectively narrow the gap with the Shannon limit. We evaluate the BER performance of IPM-16 and QAM-16 with K-means or BPS in different SNR ranges at loop = 5, as shown in Figure 5a, and the recovery constellations SNR = 16 dB are shown in Figure 5b. The BER decreases as the SNR increases, and it is evident that IPM-16 with K-means algorithm outperforms IPM without K-means and QAM-16. The hard-decision FEC (HD-FEC) threshold is 3.8 × 10 −3 , which is mainly used to evaluate the BER performance. Specifically, when SNR is small, the performance of these three schemes is approximately equal. Even the performance of QAM-16 is better than IPM at SNR = 10 to 13. With the increase in SNR, the advantages of the IPM scheme gradually appear. It is clear that IPM with K-means reach the HD-FEC threshold at SNR = 16, while the SNR of reaching the threshold of IPM without K-means and QAM-16 is 16.9 dB, 17.7 dB, respectively. Therefore, the gain of proposed schemes is 0.9 dB, 1.7 dB compared with IPM-16 without K-means cluster algorithm and QAM-16. In the diagram of recovery constellations, the black cross is the centroid locations after a serious iteration, To study the BER performance of IPM-64 and QAM-64, we draw the curves of BER versus SNR, as shown in Figure 6a. The BER gradually decreases as the SNR increases, and there is a positive correlation between BER and SNR. Additionally, it could be seen that the proposed scheme is superior to IPM-64 without K-means and QAM-64 throughout the entire range of SNR. The proposed scheme reaches the threshold of HD-FEC at SNR = 23.7 dB, while the IPM-64 without K-means cluster algorithm is at 24 dB and QAM-64 is 24.8 dB. The gap between the proposed scheme and the other two schemes is 0.3 dB, 1.1 dB, respectively. It is worth observing from the diagram of the recovery constellation that the points in the corner of QAM-64 mix together and are difficult to distinguish, while the outermost constellation points of IPM have higher discrimination, which indicates that IPM has good robustness to phase noise caused by self-phase modulation. To study the BER performance of IPM-64 and QAM-64, we draw the curves of BER versus SNR, as shown in Figure 6a. The BER gradually decreases as the SNR increases, and there is a positive correlation between BER and SNR. Additionally, it could be seen that the proposed scheme is superior to IPM-64 without K-means and QAM-64 throughout the entire range of SNR. The proposed scheme reaches the threshold of HD-FEC at SNR = 23.7 dB, while the IPM-64 without K-means cluster algorithm is at 24 dB and QAM-64 is 24.8 dB. The gap between the proposed scheme and the other two schemes is 0.3 dB, 1.1 dB, respectively. It is worth observing from the diagram of the recovery constellation that the points in the corner of QAM-64 mix together and are difficult to distinguish, while the outermost constellation points of IPM have higher discrimination, which indicates that IPM has good robustness to phase noise caused by self-phase modulation. Figure 7a shows the BER performance of the proposed scheme, IPM-256 without Kmeans and QAM-256 when loop = 1. We could observe that when SNR = 28 dB, the BER performance of QAM-256 is superior to IPM, yet not as good as the proposed scheme. When SNR > 28.8 dB, the IPM-256 without K-means has a lower BER. For comparison, our proposed scheme acquires 0.4 dB and 0.8 dB, respectively, at the HD-FEC threshold, compared with IPM-256 without K-means and QAM-256. The results fully show that although IPM-256 is not as good as QAM-256 in resisting the influence of nonlinear effect, the system's tolerance to nonlinear effect is improved after the application of the K-means algorithm. Figure 7a shows the BER performance of the proposed scheme, IPM-256 without K-means and QAM-256 when loop = 1. We could observe that when SNR = 28 dB, the BER performance of QAM-256 is superior to IPM, yet not as good as the proposed scheme. When SNR > 28.8 dB, the IPM-256 without K-means has a lower BER. For comparison, our proposed scheme acquires 0.4 dB and 0.8 dB, respectively, at the HD-FEC threshold, compared with IPM-256 without K-means and QAM-256. The results fully show that although IPM-256 is not as good as QAM-256 in resisting the influence of nonlinear effect, the system's tolerance to nonlinear effect is improved after the application of the K-means algorithm. Figure 7a shows the BER performance of the proposed scheme, IPM-256 without Kmeans and QAM-256 when loop = 1. We could observe that when SNR = 28 dB, the BER performance of QAM-256 is superior to IPM, yet not as good as the proposed scheme. When SNR > 28.8 dB, the IPM-256 without K-means has a lower BER. For comparison, our proposed scheme acquires 0.4 dB and 0.8 dB, respectively, at the HD-FEC threshold, compared with IPM-256 without K-means and QAM-256. The results fully show that although IPM-256 is not as good as QAM-256 in resisting the influence of nonlinear effect, the system's tolerance to nonlinear effect is improved after the application of the K-means Due to the involvement of NLPN and the centroid judgement problem, the signal points are excessively divergent when the algorithm converges. Consequently, the BER performance is worse than the IPM employing BPS in the initial SNR range. When employing the K-means algorithm, the SNR of the IPM scheme at HD-FEC threshold is improved dramatically compared to the other two approaches, as shown by Table 2. K-means algorithm can directly process the constellation with phase noise according to the minimum Euclidean distance criterion which will bring classification gain to system. We firstly assign M (depending on the modulation order) initial centroid positions before running the K-means algorithm and generally use the coordinates of the constellation points of the ideal constellation as the initial centroid so that the K-means algorithm can converge quickly. The new centroid of each symbol is determined by calculating the Euclidean distance between the received N symbols with phase noise and M original centroids. Then, the new centroid location coordinates are obtained based on the current constellation point clustering. Repeat the procedure above until the algorithm converges and the final centroid location is the optimum constellation point position with phase offset. Finally, each computed centroid matches the initial ideal constellation point one by one, completing the demodulation step. In contrast, BPS procedure should recover the constellation to standard firstly before demodulating it. The recovery mistake may occur if the constellation was impacted by sufficient phase noise. That is the main reason why the proposed scheme has better BER performance than other schemes. Simultaneously, the proposed technique diminishes the computing complexity by avoiding the phase recovery process. Potential Application Scope The existing OIN has a more complex network structure and network equipment, which generally complete the scheduling and management of network resources by the software defined network (SDN). The proposed scheme provides a more efficient and adaptive modulation and demodulation algorithm for SDN. SDN controller dynamically adjusts network rate, modulation order and other parameters according to the current network link characteristics and quality of service (QoS), controls the data exchange between switches, and runs network services more effectively and meets dynamic business needs. Meanwhile, the implementation does not need the corresponding demodulation and DSP module, which has lower hardware complexity and can effectively save the cost in OIN deployment. The K-means algorithm used by the receiver is also apt to integrate with SDN, which provides theoretical support for two-tier OIN. In addition, we all know that the average communication distance between data centers in the same city, especially disaster-tolerant backup data center, is about 60 km, and the longest distance does not exceed 200 km. However, we simulated the MI of each scheme under different loops in Section 3. The maximum loop is equal to 10, reaching 1000 km. At the same time, the proposed scheme shows better BER performance when the modulation formats are 16-IPM and 64-IPM and the simulated transmission distance is 500 km. Since the simulated transmission distance exceeds the actual distance between data centers, and the proposed scheme has better performance, which jointly decide that it can be used not only for OINs, but also for any high-speed and long-distance optical fiber communication system. Conclusions In conclusion, the state-of-the-art scheme that applied K-means cluster algorithm in geometric shaping based on iterative polar modulation has been proposed, which could be employed in OINs. Even if the transmission distance is sufficient, the proposed scheme could perform better when K-means cluster algorithm in IPM is used, which opens up new applications for it in high-speed, long-distance optical fiber communication systems. We establish a 60 Gbps coherent optical transmission simulation system to verify the performance of our proposal. In terms of mutual information, whether in long/shortdistance transmission, IPM could provide a large capacity for the transmission system. For the BER performance, the scheme of IPM-16 with K-means achieves 0.9 dB and 1.7 dB, respectively, compared with IPM-16 without K-means and QAM-16; the scheme of IPM-64 with K-means achieves 0.3 dB and 1.1 dB, respectively, compared with IPM without K-means and QAM 64; the scheme of IPM-256 achieves 0.4 dB, and 0.8 dB, respectively, compared with IPM 256 without K-means and QAM-256. The scheme of IPM with Kmeans is superior to the other two schemes, which behave as strong robustness to NLPN, according to the aforementioned results, whether in terms of BER or channel capacity. We consider that our proposed scheme will provide a new idea for developing transmission technology of inter-data center optical interconnection networks.	2021-10-18T17:16:21.337Z	2021-10-03T00:00:00.000	{ "year": 2021, "sha1": "c7870379eb0caec18ab1a2db384cc6c5ec507eb1", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2079-9292/10/19/2417/pdf", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "ef05fbbee5dfadcf004b3fcb768df59fc349e75c", "s2fieldsofstudy": [ "Engineering", "Physics", "Computer Science" ], "extfieldsofstudy": [ "Computer Science" ] }
245954541	pes2o/s2orc	v3-fos-license	Robust predictive Control for Networked Control Systems with Access Constraints : Based on the rolling horizon optimization strategy, the networked robust predictive control with medium access constraints and packet loss is studied. Firstly, considering the influence of network factors such as medium access constraints and packet loss, Markov jump rule and Bernoulli independent and identically distributed process are used to transform the network problem into the robust problem of networked control system. According to the established Markov jump system model and stability analysis, a robust predictive controller for networked control systems is designed by using linear matrix inequality (LMI) method, which makes the system asymptotically stable. Finally, a numerical example is given to verify the effectiveness of the proposed control method.Networked control system; access packet controller. convenience, networked control system also has some problems that can not be ignored. The problems of media access constraints and packet loss in networked control system are also very common. Media access constraints have a great impact on system stability. The so-called media access constraints are that because the bandwidth of each network is limited, at any time, only a few sensors and actuators will obtain channel access rights for data transmission and exchange [4]. People have conducted in-depth research on the problems caused by media access constraints. The authors of [5] studies the linear steady network control problem with medium access constraints. The system is modeled as a switched system, and then the sufficient conditions for the stability of the system are given on the basis of dwell time by using Lyapunov function method. The authors of [6] studies the minimum data rate problem of networked control systems under media access constraints. In order to make the system stable, a new analysis method of unstable scalar and vector is established, and finally the necessary and sufficient conditions for system stability are obtained. The authors of [7] studies the second-order multi-agent formation problem of wireless networks under media access constraints. By designing the agent node scheduling protocol, the qualified agent nodes can transmit information through the network. The authors of [8] studies the event triggered collaborative design method of nonlinear networked control systems with medium access constraints. On the basis of considering nonlinear dynamics, an event triggered scheme with adjustable trigger conditions is proposed. For networked control systems with media access constraints, different analysis methods are used to make the system asymptotically stable, but only a single media access constraint problem is considered in the above literature, which is not practical in practice. Packet loss is also one of the common problems in networked control system. Packet loss is a phenomenon that the transmitted data cannot reach or reach the next node in time based on the reliability of the network itself and the influence of the network environment [9]. For some control systems with robust performance, although a certain probability of packet loss can be allowed, it will also have a certain impact on the system performance, which may lead to system instability. The impact of packet loss on system performance is also an important factor worthy of attention. Under this background, many scholars have conducted in-depth research on packet loss. Aiming at the problem of random packet loss, literature [10] estimates the state of networked control system, and eliminates various effects of packet loss on system state estimation by solving linear matrix. The authors of [11] studies the simultaneous estimation of equipment state and packet loss at each time step. After solving the problem, two solutions are proposed. Finally, an example is given to verify the effectiveness of the proposed method. The authors of [12] studies the problem of fault detection in nonlinear networked control systems. Assuming that data packets are lost during data transmission, it is modeled as a white noise sequence with Bernoulli distribution. The authors of [13] studies the dynamic output feedback fuzzy control problem triggered by adaptive events. The data packet loss is described by Bernoulli random process, and a fuzzy dynamic output feedback controller is designed to ensure the stability of the system. The above literature designs the controller through different methods for the networked control system with packet loss, so that the system can reach an asymptotically stable state in case of network packet loss. However, in reality, media access constraints and packet loss usually exist at the same time. The above control method for a single network problem is relatively conservative. Now most research work is based on an ideal assumption, that is, the system noise and process noise meet the white noise of Bernoulli distribution. In the actual industrial process, the noise is not the white noise that simply meets the characteristics of Bernoulli distribution, but some information sources with limited energy. If we ignore these practical useful information, it will inevitably reduce the estimation accuracy and estimation performance. However, most literatures only consider the influence of a single factor, which makes the system performance not optimal. In this paper, considering the influence of network factors such as medium access constraints and packet loss, the network problem is transformed into the robustness problem of networked control system by using Markov jump rule and Bernoulli independent identically distributed process. According to the established Markov jump system model and stability analysis, a robust predictive controller for networked control system is designed by using linear matrix inequality method, so that the system can reach asymptotic stability. Considering the NCS shown in Figure 1, the linear discrete system model of the controlled object is ( ) Model establishment where n x R ∈ is the state vector; m u R ∈ is the input vectors, respectively, , , A B D representing constant matrices of appropriate dimensions. In addition, it is assumed that the system noise is not white noise, but regarded as an unknown bounded deterministic variable. The NCS in Figure 1 has m actuators and n sensors. Information is exchanged between the controller and the actuator through the network, and there are no communication constraints between the controller and the sensor. Due to the limitation of network bandwidth, only p (1 ) p m ≤ < actuator can obtain the communication channel in each sampling period and execute the control command transmitted by the controller, therefore, there are p m m N C = media access modes for the actuator end of NCS. In addition, control packet loss may occur randomly during transmission due to node failure, network congestion or communication noise. In order to describe the media access state of the actuator, the following binary functions are defined ( ) , then the input of the controlled object is Assuming that the switching between two adjacent sampling periods it is assumed that the control rate of the system is given by state feedback For the packet loss problem, we introduce a binary random variable ( ), 1, , to represent the media access state of the packet loss process at time k . When the packet is successfully transmitted to the actuator ( ) 1 i k σ = , when the packet is lost during transmission ( ) 0 i k σ = . It is assumed that for the case of any packet loss, it conforms to the Bernoulli random process of independent and identically distributed, and has the following probability distribution: is a known constant, which represents the arrival probability of the i channel packet of the control input. Considering the impact of packet loss, the final control signal received from the actuator can be expressed as follows: represents the probability matrix of successful transmission of control signal. Based on equations (1), (2) and (6), the discrete model of networked control system can be described as Stability analysis Before the analysis of the main results, the following lemmas are given: Lemma1 [14] : Assumed symmetric matrix , where , then the following conditions are equivalent: , . Lemma2 [14] : , H E is a real matrix with appropriate dimension, F is a positive definite symmetric matrix and satisfies T F F I ≤ , then there is a scalar 0 ε > such that the following inequality holds: As we all know, model predictive control has the advantages of good control effect and strong robustness. It can deal with various constraints in controlled variables and operating variables. Firstly, the following prediction model is given: represents the controller output of predicting the k l + time based on the known information of k time, then the prediction output sequence 0 U ∞ of the controller can be expressed by the following formula: Among them, ( ) where the symmetry matrix x k l k P s x k l k x k l k P i x k l k the following robust stability constraints are satisfied: under the condition that the system (7) is asymptotically stable, ( ) , the optimization problem of robust predictive control for system (7) can be further described as: and meet the constraints (13) and mandatory constraints Theorem 1: Assumes that the state of networked control system (7) then the networked control system (7) is asymptotically stable and has certain control Proof:Firstly, a sufficient condition for the robust stability of networked control system (7) is proved, so that inequality (18) holds. It can be obtained from equation (8), equation (12) and equation (13) inequality (21) can be obtained by applying lemma 1 to inequality (20), inequality (21) can be decomposed into inequality (23) can be obtained by applying lemma 2 to the inequality (22) obviously, if the inequality (24) holds: then inequality (18) can be proved by lemma 1. Next, it is proved that inequality (19) holds, system (7) has certain control performance, inequality (16) can be further transformed into inequality (26) can be obtained by applying lemma 1 to inequality (25) where ( ) P i , ( ) P j is a nonsingular matrix. Contract transformation is performed on equation (34), that is, the left and right sides are multiplied by the diagonal matrix at the same time inequality (35) can be rewritten as of Markov communication sequence between controller and actuator, and "1" and "2" in the ordinate represent the random matrix of communication sequence 1 M and 2 M . figure.4 shows the packet loss state of the system, which "0" and "1" in the ordinate represent the packet loss state and no packet loss respectively. figure. 5 is the input signal of the controller. It can be seen from the figure that the system can quickly reach the asymptotically stable state in the 10th sampling period. Figure 6 is the state trajectory diagram of the system. It can be seen from the diagram that under given initial conditions, the controller designed in this paper can quickly converge the state of the closed-loop system (7) to an asymptotically stable state at the 10th sampling period. From the simulation results, we can see that the response fluctuation of the system is small, which shows that the networked robust predictive controller designed in this paper can make the closed-loop system quickly reach asymptotic stability and has good dynamic performance. Conclusions Based on the rolling time domain optimization strategy, this paper studies the networked robust predictive control with medium access constraints and packet loss. Considering the influence of network factors such as medium access constraints and packet loss, the network problem is transformed into the robustness problem of networked control system by using Markov jump rule and Bernoulli independent identically distributed process. According to the established Markov jump system model and stability analysis, a robust predictive controller for networked control system is designed by using linear matrix inequality method, so that the system can reach asymptotic stability. Finally, a numerical example is given to verify the effectiveness of the proposed control method.	2022-01-15T16:24:39.646Z	2022-01-01T00:00:00.000	{ "year": 2022, "sha1": "abcd7a49e3bd92185f0243a00760c8455fb57e25", "oa_license": "CCBY", "oa_url": "http://www.clausiuspress.com/assets/default/article/2022/01/12/article_1642033770.pdf", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "d568f39c0805ab6c4860c3e689840b0f424c1ca1", "s2fieldsofstudy": [ "Engineering", "Computer Science" ], "extfieldsofstudy": [] }
41481849	pes2o/s2orc	v3-fos-license	Pentoxifylline as Adjuvant Therapy to Etanercept in Patients with Moderately to Highly Active Rheumatoid Arthritis Objective: Rheumatoid arthritis (RA) is a common inflammatory disease associated with many extraarticular features. The aim of the study was to evaluate the effects of pentoxifylline (PTX) as adjuvant therapy to etanercept in moderately to highly active RA. Methods: A single center randomized double-blind placebocontrolled trial of 8 weeks duration was performed. Disease activity was measured via calculating the disease activity score in 28 joints using erythrocyte sedimentation rate (DAS28-ESR) and by simplified disease activity score in 28 joints using hsCRP (SDAI-CRP). 40 Patients who were using etanercept (ETN) were randomly allocated to receive each day either pentoxifylline 400mg tablet twice daily or capsules prefilled with glucose as placebo also twice daily and were evaluated at baseline and at week 8 for clinical and hematological parameters. Results: Tumor necrosis factor (TNF), high sensitive C-reactive protein (hsCRP), duration of morning stiffness, and cardiovascular risk were significantly more reduced in pentoxifylline group than placebo group after 8weeks. Non significant changes were observed in clinical parameter like swelling joints counts (SJC),tender joints counts(TJC),visual analogue scale(VAS),evaluator global assessment (EGA),DAS28-ESR, SDAI-CRP and hematological parameter like hemoglobin (Hb) amount, erythrocyte sedimentation rate (ESR) and white blood cells (WBC) count between groups. Conclusion: PTX significantly decreased pro-inflammatory markers (TNF, hsCRP), duration of morning stiffness and cardiovascular risk. This suggests that pentoxifylline may be a promising and useful strategy to reduce the systemic inflammation and cardiovascular morbidity and mortality observed in RA patients. Introduction Rheumatoid arthritis (RA) is a chronic systemic autoimmune inflammatory disease that affects all ethnic groups throughout the world [1] with increased morbidity and mortality from premature cardiovascular disease (CVD). Up to 50% of this excess mortality is secondary to ischemic heart disease (IHD) closely followed by cerebrovascular disease [2], with a 1.5-fold increase in the standardized mortality ratio due to CV events compared with the general population [3]. The principle inflammatory cytokines that believed to have a key position in the pathogenesis of RA include TNF-α and IL-6, both of which are targets for therapy [4]. Pentoxifylline is a tri-substituted xanthine derivative designated chemically as 3, 7-Dihydro-3, 7-dimethyl-1-(5oxohexyl)-1H-purine-2, 6-dione that, unlike theophylline, is a hemorheologic agent, i.e., an agent that affects blood viscosity [5]. Pentoxifylline is frequently prescribed for patients with peripheral arterial disease, mainly because of its capacity to deform hemocytes and to its vasodilatation effect [6]. This drug can also decrease levels of tumoral necrosis factor, an important inflammatory cytokine responsible for the increased endothelial expression of adhesion molecules and for IL-6 production [7]. The precise mode of action of pentoxifylline and the sequence of events leading to clinical improvement are still to be defined [5]. Many studies have demonstrated the possible antiinflammatory effect of pentoxifylline (PTX). A prospective 3-month open evaluation of PTX in a group of adult patients with RA refractory to conventional disease curative therapies in 1995 where 19 patients with RA assessed clinically according to the World Health Organization/International League of Associations for Rheumatology (WHO/ILAR) criteria at baseline, and 3 months after the initiation of therapy. They evaluated a complete blood count, erythrocyte sedimentation rate (ESR), and whole blood assays of TNF-α production. After 3 months they found a significant diminution in number of tender and swollen joints as well as the ESR (p < 0.05) although no consistent effects on TNF-α production were observed. Furthermore, whole blood assays of TNF-alpha production shortly after initiation of pentoxifylline therapy were not predictive of the clinical response to this agent and they concluded that although pentoxifylline may possess therapeutic properties in RA, any beneficial effects appear to be unrelated to changes in TNF-α generation in whole blood assays [8]. Also Dysregulation of TNF-α production which thought to be important in rheumatoid arthritis was tested in 1995 by using combination of PTX and thalidomide in an open study of rheumatoid arthritis patients to assess the effect on TNF production, the antiarthritic effects and toxicity of this combination, in which 12 patients with active rheumatoid arthritis were treated with 1200 mg pentoxifylline and 100 mg thalidomide daily for 12 weeks. The results showed a lowering in the TNF production capacity during treatment (P < 0.03) whereas production capacity of IL-6, IL-10, and IL-12 was not affected. They concluded that although pentoxifylline/thalidomide reduced the production capacity of TNF, the benefit/side effects ratio was poor due to multiple adverse effects, while clinical observation suggests limited efficacy [9]. Another study on a 64 year-old man with seronegative RA who had 23 swollen joints, 32 painful joints, and ESR 135 mm/h. All these parameters were dramatically improved 3 weeks after administration of PTX 300 mg/d and prednisolone 5 mg/d. Discontinuation of PTX in this patient resulted in rapid exacerbation of RA and when PTX was restarted the patient showed complete recovery from arthritis with normalization of ESR within 3 months and was maintained a complete remission for another 1 year. This case further supports a potential antirheumatic effect of PTX on some patients with RA [10]. Also, in an animal model, Al-Saad et al evaluated the dose-response relationship of the anti-inflammatory activity of PTX in experimental animal models of chronic inflammation in 2012 and found that PTX in a dose dependent pattern attenuates formaldehyde-induced chronic inflammation and cotton-pellet induced granuloma in rats and potentiates the anti-inflammatory activity of dexamethasone and methotrexate [11]. The aim of this study was to evaluate the effects of PTX as adjuvant therapy to etanercept in moderately to highly active RA. Study Design This was an 8-week randomized double blind placebo controlled single center trial conducted at Rheumatology Unit, Baghdad Teaching Hospital, Baghdad, Iraq carried out over 5 months from September 2012 till March 2012 at Rheumatology Unit, Baghdad Teaching Hospital. Patients were randomly allocated to receive each day either pentoxifylline400 mg tablet twice daily or capsules prefilled with glucose as placebo (PBO) orally also twice daily in addition to their weekly subcutaneous etanercept dose. Pentoxifylline was bought from Sanofi-Aventis-Egypt Company, Egypt. Whereas glucose was bought from SDI, Samarra, Iraq. Patients were evaluated at baseline and at week 8. Informed consent was obtained from all participants and this study was approved by the ethical committee of Baghdad University, College of Medicine -Medical Department. Sample Selection Eligible patients had confirmed RA according to the 1987 American College of Rheumatology (ACR) criteria with moderate to highly active disease defined as disease activity score based on 28 joints and ESR (DAS28-ESR) greater than 3.2 at baseline. For inclusion, patients also were required to have taken etanercept subcutaneously and regularly for at least 4 previous consecutive months without any clinical response. The exclusion criteria included presence of contraindication to PTX (Cerebral hemorrhage, extensive retinal hemorrhage, acute myocardial infaction, and sever cardiac arrhythmia), renalimpairment, hepatic impairment, pregnancy, breast feeding, patients already onPTX, patients with mild or inactive RA,using DMARDs other than etanercept , with high dose of NSAIDS, and patients with co-existent other connective tissue disease. Additionally, 20 healthy age and sex matched individuals were considered as a control group. Clinical and Laboratory Evaluation Clinical evaluation of patients for tender and swelling joints was done by specialized rheumatologist who was blinded to treatment at zero time (baseline) and after 8 weeks. The RA disease activity was measured using DAS28-ESR) [12] and Simplified Disease Activity Index (SDAI) [13]. DAS28 is calculated from the number of tender and swollen joint counts (TJC and SJC; 28-joint count), patient self-assessment of disease activity (visual analog scale, VAS), and erythrocyte sedimentation rate (ESR.]. SDAI is the numerical sum of five outcome parameters: TJC and SJC (based on a 28-joint assessment), patient and physician global assessment of disease activity (VAS 0-10 cm) and level of C reactive protein (CRP).DAS28 and SDAI can be calculated according to the following formula: DAS28 = 0.56 (TJC) 0.5+0.28(SJC) 0.5+0.70ln (ESR) +0.014(VAS) SDAI=TJC+SJC+EGA+VAS+CRP Blood specimen collection and laboratory analysis (at baseline and after 8 weeks) of WBCs count, ESR and Hb, high sensitive CRP (hsCRP),TNF-α, was done by specialized laboratory researchers who did not participate in this study. WBCs count and Hb were measured by a hematology auto-analyzer (Ruby-CELL-DYN 08H56-02, Abbott Company, Abbott Park, IL, USA). ESR was measured by Westergren method [10eh].hsCRP [14] and TNF-α [15] by using ELIZA technique. Statistical Analysis Statistical software (SPSS version 16, Chicago, IL, USA) was used for data input and analysis. Continuous variables were presented as mean ± standard deviation (SD) and discrete variables were presented as numbers and frequencies. Chi square test for independence was used to test the significance of association between discrete variables. Continuous variables were tested by the Shapiro Wilk test to determine if they were normally or abnormally distributed. ANOVA test was used to test the significance of difference in the mean of 3 independent samples in normally distributed continuous variables. Unpaired t-test was used to test the significance of difference in the mean of two independent samples in normally distributed continuous variables and Mann Whitney test for abnormally distributed data. Findings with P value less than0.05 were considered significant whereas P values less than 0.01 considered highly significant. Results Of a total of 49 patients who were randomized in this double-blind study, 40 completed the 8 weeks of treatment (20 from the pentoxifylline (PTX) group and 20 from the placebo) (Figure 1). The mean age of PTX group did not differ significantly from placebo and controls(50.25±9.91years vs 49.8 ±9.84years vs 42.05±9.88 years, P>0.05) respectively. Also there was no statistical significant difference in the female to male ratio among PTX group, placebo group, and controls (15:5 (75%)vs12:8 (60%) vs10:10(50%), P>0.05) respectively (Table 1). Baseline hematological parameters showed that Hb level was significantly lower in patients with active RA than healthy control subjects (P < 0.05).Whereas ESR and WBC count were significantly higher in patients with active RA than those in control group (P < 0.001, Table 2). Table 3 showed no significant difference in the baseline clinical parameters between PTX and placebo group patients (P>0.05). After 8 weeks of starting adjuvant treatment with eitherPTX or placebo, we found that only hsCRP and TNF-α decreased significantly by PTX (P < 0.05, Table 4). In addition, only duration of morning stiffness decreased significantly by PTX (P < 0.05) while other parameters showed no significant difference between the effect of PTX and placebo (P > 0.05, Table 5). Discussion Swelling joints are considered as the best single variable that detects response to drug therapy in RA patients [16]. The present study showed that PTX reduced swelling joints significantly while placebo did not produce any significant change after 8weeks of therapy, but overall there was non significant difference between the effect of PTX and placebo on SJC. This finding was different from a randomized placebo controlled trial that used PTX 400mg tablets three times daily or placebo for 24 weeks which showed that PTX was more effective than placebo in term of reducing SJC [17]. This difference can be attributed to the short duration of this study in comparison to the above study that continued for 24 weeks. Assessment of tender points is the cornerstone during evaluation and treatment decision making in RA [18]. This study showed that there was a non-significant difference between the effect of PTX and placebo; similar finding was reported by Maksymowych et al [19] who carried out a prospective 3-month open evaluation of PTX in a group of adult patients with RA refractory to conventional disease remittive therapies and found PTX was effective to reduce TJC in patients with mild active RA more than placebo but the effect was not statistically significant. Visual analogue scale (VAS) is a valid measure of pain intensity in RA patients [20]. The current study showed that there was a non-significant difference between the effect of PTX and placebo on VAS. Similar study in which the effect of PTX on painful distal diabetic neuropathy was measured by Cohen et al who used PTX in six-month trial for the treatment of painful distal diabetic neuropathy and found non-significant difference in the patients' pain between the PTX-and placebo-treated groups [21]. But Usha et al [17] reported that PTX reduced VAS significantly while placebo not. The difference may be due to the difference in duration of the present study or the use of etanercept with placebo that were also effective in reducing VAS and may cover the effect of PTX. Evaluator global assessment (EGA) is considered a reliable indicator for assessing RA and even more reliable than VAS [22]. We found a non-significant difference between the effect of PTX and placebo on EGA. Exactly there were no similar studies in this respect and this was the first study evaluated the effect of PTX on EGA. Acute phase reactants ESR and CRP provide reliable means for discrimination between drugs that provide symptomatic relief only and those with a more profound effect in RA [23]. In this study there was non-significant difference of the effect of PTX on ESR. In contrast to Maksymowych [19] who found PTX decreased ESR significantly after 3 months. Possible explanation may be the effect of etanercept in our patients making effect of PTX much harder to detect. The present study showed that RA patients had a significantly higher CRP level than healthy control subjects, which was similar to another study that reported a higher CRP level in RA patients than in healthy control subjects and CRP elevation was directly correlated with RA disease activity [24]. The results also showed a significant effect of PTX but not placebo on hsCRP level this result was similar to randomized placebo-controlled study done by Fernande et al [25] who measured proinflammatory and anti-inflammatory activity of PTX in patients with coronary artery disease and concluded that PTX reduced pro-inflammatory CRP level significantly. Another study done by Maitia et al [26] showed after onemonth follow-up PTX decreased significantly C-reactive protein and ESR. Regarding RA disease activity, PTX had non significant effect on both DAS28 and SDAI. This was the first study that measured the effect of PTX on DAS28 or SDAI and the result can be explained since these measures depend on different factors include TJC, SJC, VAS, ESR, hsCRP so the effect of PTX will be the result of effect on the above factors which showed a high percent of change in PTX group but non significant effect. The results of this study showed a significant difference between the effect of PTX and placebo on TNF-α. Similarly González-Espinoza et al [27] showed that PTX significantly decreased serum concentrations of TNF-α, IL-6 and CRP compared to placebo. This may suggest that PTX could be a promising and useful strategy to reduce the systemic inflammation. Additionally, we found that PTX reduced morning stiffness significantly. Inconsistent with Huizinga et al [28] who found that PTX produced non-significant change in morning stiffness when used in combination with thalidomide. The finding in our study was expected possibly because PTX could improve the markers of inflammation that morning stiffness was correlated with like ESR, swollen joint counts, pain, fatigue, tender joint and patient and physician global assessment of disease activity [29]. Also the above study included only 12 patients and used PTX as adjuvant with thalidomide which differs in mechanism of action and efficacy from etanercept which was used in current study. The results of this study showed that Hb level in RA patients was significantly less than control subjects, this finding was proved by many other studies showing that anemia was a common extra articular manifestation of RA patients [30,31]. Also neither PTX nor placebo could improve Hb level which was similar to another animal study done by Mendes et al who found PTX decreased hemoglobin content by approximately 40% [32]. Moreover, WBC count was significantly higher in RA patients than control subjects; this finding was similar to another clinical trial that showed a high level of WBC count in RA patients [33] whereas other studies found a correlation between inflammation and WBC count [34]. The data also showed that PTX was no more effective than placebo to decrease WBC count, Inconsistent with animal study done by Mohammadzadeh et al [35] who concluded PTX could decrease total white blood cell count in serum of rats. Furthermore, PTX produced a significant change in hsCRP and TNF which are the most important markers for CV risk specially the level of circulating CRP which is a prognostic marker of future CV events for men and women both with and without established CVD [36].The non-significant effect on other inflammatory parameters that increase cardiovascular risk like ESR may be related to use of other drugs with PTX and placebo that may mask the positive effect of PTX on these CV risk factors. Fernandesa, et al [37] reported PTX reduced the proinflammatory and increased anti-inflammatory response in patients with acute coronary syndrome and may have beneficial clinical effects on cardiovascular events. In conclusion, PTX significantly decreased proinflammatory markers (TNF, hsCRP), duration of morning stiffness, and cardiovascular risk. This may suggest that PTX could be a promising and useful strategy to reduce the systemic inflammation and cardiovascular morbidity and mortality observed in RA patients.	2019-03-11T13:12:12.998Z	2013-01-23T00:00:00.000	{ "year": 2013, "sha1": "6a4b91699dd541ea282eff45386cb655a78b1c32", "oa_license": null, "oa_url": "http://pubs.sciepub.com/ajps/1/4/4/ajps-1-4-4.pdf", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "ef0a0a074eae5be0c628545d26a7ad7379117a3b", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
2577362	pes2o/s2orc	v3-fos-license	Assessing Women’s Preferences and Preference Modeling for Breast Reconstruction Decision Making Background: Women considering breast reconstruction must make challenging trade-offs among issues that often conflict. It may be useful to quantify possible outcomes using a single summary measure to aid a breast cancer patient in choosing a form of breast reconstruction. Methods: In this study, we used multiattribute utility theory to combine multiple objectives to yield a summary value using 9 different preference models. We elicited the preferences of 36 women, aged 32 or older with no history of breast cancer, for the patient-reported outcome measures of breast satisfaction, psychosocial well-being, chest well-being, abdominal well-being, and sexual well-being as measured by the BREAST-Q in addition to time lost to reconstruction and out-of-pocket cost. Participants ranked hypothetical breast reconstruction outcomes. We examined each multiattribute utility preference model and assessed how often each model agreed with participants’ rankings. Results: The median amount of time required to assess preferences was 34 minutes. Agreement among the 9 preference models with the participants ranged from 75.9% to 78.9%. None of the preference models performed significantly worse than the best-performing risk-averse multiplicative model. We hypothesize an average theoretical agreement of 94.6% for this model if participant error is included. There was a statistically significant positive correlation with more unequal distribution of weight given to the 7 attributes. Conclusions: We recommend the risk-averse multiplicative model for modeling the preferences of patients considering different forms of breast reconstruction because it agreed most often with the participants in this study. the probabilities that those outcomes will occur. Furthermore, she must evaluate the value of each outcome. However, the number of possible breast reconstruction outcomes is large, 1 thereby making it infeasible to evaluate each possibility. Most physicians spend between 46 and 60 minutes during the initial consult alone. 4 Multiattribute utility theory (MAUT) 5,6 enables the quantification of outcomes given the preferences of the patient. This quantification can then be used to assist with medical decision making. We are unaware of any studies that apply this theory in breast reconstruction decision making; however, it has been applied in other areas of healthcare such as prostate cancer treatment and neonatal intensive care. [7][8][9][10][11][12][13] Yet, in all cases, MAUT was applied without evaluating its performance in modeling preferences. If MAUT accurately models women's preferences in the breast reconstruction decisionmaking setting, it may be employed by future computational decision support systems to aid in patient decision making about breast reconstruction. Although this study was conducted on healthy participants, the findings may be extended to all patients who have not been informed of a diagnosis of breast cancer. The application of MAUT may enable plastic surgeons to quickly and efficiently identify breast reconstruction choices that are more preferred by the patient. Defining an Outcome: The 7 Attributes An outcome is composed of a set of attributes and associated values. The BREAST-Q is a validated and commonly employed instrument for assessing satisfaction and quality-of-life outcomes of breast reconstruction. The BREAST-Q assesses 5 patientreported outcome measures (or attributes): (1) satisfaction with one's breasts, (2) psychological well-being, (3) well-being of the chest, (4) well-being of the abdomen, and (5) sexual well-being. [14][15][16][17] For each measure, the BREAST-Q yields a raw point total, which may be converted into a score (between 0 and 100) using a propriety scoring algorithm. The BREAST-Q has a number of breast reconstruction variants. As the participants in this study have never had breast cancer or mastectomy, we deployed the preoperative version. In addition to the BREAST-Q measures, we considered the attributes of (6) cost in terms of out-ofpocket expenses and (7) time lost to reconstruction or associated medical procedures. Out-of-pocket cost is measured in US dollars and time is measured in days. Participants The study investigated the breast reconstruction outcome preferences of 36 women between May 1 and July 18, 2013, in the Seattle, WA, Austin, TX, and Houston, TX, areas. Participants in this study were healthy human volunteers who did not have a history of breast cancer. We obtained Institutional Review Board approval from The University of Texas MD Anderson Cancer Center to conduct this research. The protocol title is recorded as "Validating a Multiattribute Utility Function for Breast Reconstruction Decision-Making." The MD Anderson protocol ID number is PA12-1179, approved on February 28, 2013. A similar protocol was separately submitted to the Institutional Review Board at The University of Texas at Austin, which determined that it was not human subjects research because no identifying or health information was to be collected. Structured Preference Elicitation Consultation Individual consultations were conducted in person with the first author. During the structured consultation, the participant performed a series of tasks: (1) assess BREAST-Q state and clarified attributes, (2) assess risk tolerance for cost and time, (3) clarify preferences for all 7 attributes, and (4) rank 10 sets of 3 standardized hypothetical breast reconstruction outcomes. Evaluating Preference Model Performance We quantified the performance of each preference model in terms of its consistency, correctness resolution, and error resolution. The consistency of a model is the extent to which its rankings of the breast reconstruction outcomes matched the rankings of the participants. At the end of the consultation, we asked each participant to rank 10 sets of randomly generated hypothetical breast reconstruction outcomes in order of preference from best to worst. When generating these outcomes, we limited the possible values to between the best and worst values as given in Table 1. Each set contained 3 outcomes (Fig. 1). 18 The participants were not shown the model rankings to avoid bias. Within each set, we compared the rankings of each pair of outcomes, 3 pairs per set for a total of 30 pairs. Ideally, a model would be 100% consistent with the participant, meaning that the model agreed with all the participant's rankings. By chance alone, one would expect a consistency of 50%. Hence, any consistency greater than 50% demonstrates some ability to accurately model a participant's preferences. A model that is more consistent better represents patients' preferences about breast reconstruction and produces rankings that are more likely be useful for supporting patient decision making. Each preference model expresses the value of an outcome as a u-value, which is a number between 1 and 0. The higher the u-value, the better the outcome. The u-value may be thought of as a preference probability. The correctness resolution is the u-value (probability of preference) difference with which a model correctly differentiates 2 outcomes. For instance, if a model correctly orders 2 outcomes with u-values of 0.30 and 0.35, the model has a correctness resolution of 0.05. A correctness resolution of <0.03 is preferred because this amounts to less than a 3% difference in outcomes for which people are usually indifferent. 19,20 The error resolution is the u-value difference with which a model incorrectly differentiates 2 outcomes. In other words, if a model incorrectly orders 2 outcomes of 0.30 and 0.35, then the error resolution is 0.05. The error resolution should ideally be less than or equal to the correctness resolution. Assessing Preference Weights for Attributes An individual may weigh one attribute more or less heavily than the others. For example, a breast cancer patient who has substantial caregiver responsibilities for family members might value time over out-ofpocket cost while a patient with fewer such demands on her time might value cost over time. We assessed the preference weight, k, for each attribute, using the von Neumann-Morgenstern standard gamble. 21 The bigger the k, the more preferred the attribute. To eliminate the factor of numeracy and reduce the effect of cognitive biases such as anchoring and adjustment 22 in obtaining these k values, we employed a probability wheel (AnalyCorp, Stanford, Calif.), which presents the probability as an area of a pie chart slice rather than numerical value. 19,23 To quantify a participant's preference weight distribution, we calculated where the ks are normalized such that they sum to one. 24 For instance, if the participant places equal weight on cost, time, satisfaction, psychological well-being, chest well-being, abdominal well-being, and sexual well-being, the entropy will equal a minimum of -1.95. If a participant places all the weight on one attribute only, such as satisfaction with her breasts, the entropy will equal a maximum of zero. Modeling Patient Preferences with Multiattribute Utility Theory A preference model includes 2 components: (1) the risk model for each attribute and (2) the multiattribute utility function that performs trade-offs among attributes and calculates the summary value. We considered 3 risk models and 3 multiattribute utility functions. We integrated the patient's risk attitude and preferences into 9 preference models. Risk Models Attribute values such as time and cost must first be converted into a u-value. We considered 3 types of risk models: (1) risk neutral, (2) risk averse, and (3) risk averse-preferring or sigmoidal (Fig. 2). For instance, with respect to out-of-pocket reconstruction cost, a risk neutral participant would place a value of $1000 on a 50-50 chance of a reconstruction that costs $2000 and a reconstruction that costs $0. A risk-averse participant would value the same gamble at more than $1000. A riskpreferring participant would value the gamble at less than $1000. The risk neutral model is simply a straight line representing the expected value or average. For the risk-averse modeling of out-of-pocket cost, we chose the exponential curve u x a be where x is some cost value, a and b are chosen such that u cost (x worst ) = 0 and u cost (x best ) = 1, and the "perceived monetary wealth" is obtained by determining a maximum value, x, such that the decision maker is indifferent between receiving nothing or a 50% chance at x and 50% chance at -x/2. 25 Similarly, the risk-averse modeling of time used the same exponential curve As we know the participants' BREAST-Q states, we chose the logarithmic curve for the satisfaction, psychological well-being, chest well-being, abdominal well-being, and sexual well-being attributes. For instance, for psychological well-being, the model was . 1. three randomly generated hypothetical breast reconstruction outcomes (or "situations") that participants were asked to rank in order of preference during the consultation. all participants were given the same set of situations. an outcome is composed of a set of attributes and associated values. Participants were also provided with a corresponding set of written descriptions for each outcome. attributes were color-coded to improve user-friendliness. note that the BreaSt-Q "chest WB" and "abs WB" point system is negatively oriented (less is better), whereas the colored bars are all positively oriented (more is better). As an alternative to the risk averse and risk neutral models, we also considered a hybrid risk averse-risk preferring or sigmoidal model for the satisfaction, psychological, chest, abdominal, and sexual wellbeing attributes. The BREAST-Q propriety scoring algorithm provides this risk model by producing a score between 0 and 100, given the raw point total of these attributes. We treated the BREAST-Q scores, as percentages, as the u-values. Multiattribute Utility Functions There are several candidate multiattribute utility functions that utilize participant preferences for attributes to produce an overall u-value for a given outcome. This overall u-value is, in effect, a single summary value for that outcome. Due to the number of attributes, it was only feasible to evaluate 3 functions because others require exponentially more weight assessments (eg, the multilinear requires 126). These functions were the multiplicative, 5,6,11 additive, 5,6,11,19,26 and power additive. 27 Each would require assessment of 7 preference weights. The power additive function requires an eighth risk tolerance assessment. 27 RESULTS For the 36 participants, the mean age was 51.2 years with a range of 32 to 72. With regard to race, 66.7% were White, 30.6% were Asian, and 2.8% were Black. One White participant was Hispanic (2.8%). Regarding the highest level of completed education, one participant (2.8%) graduated from high school, 4 (11.1%) had a 2-year college degree, 20 (55.6%) had a 4-year college degree, 9 (25%) had a master's degree, and 2 (5.6%) had a doctoral degree. The risk attitude of participants for out-of-pocket cost and time varied widely between risk neutral and extreme risk aversion (Figs. 3 and 4, respectively). Risk aversion was strongest and more common with respect to cost than time. Participants had a wide variety of BREAST-Q scores and preferences for attributes. The median time required to complete all sections of the consultation was 46 minutes, though with a wide range from as low as 23 minutes to as long as 127 minutes ( Table 2). We examined the consistency of the 9 combinations of risk models and multiattribute utility functions (Table 3 and Fig. 5). The best-performing preference model was the risk-averse multiplicative model with an average consistency of 78.9%. Although consistencies were lower for the remaining models, particularly the sigmoidal power additive model, they were not significantly different using the Wilcoxon signed-rank test. We achieved a power ≥0.80 for most of the models, except for the risk neutral models. However, the risk neutral multiplicative (post hoc power = 0.78) required 78 samples and the additive and power additive (post hoc power = 0.43) each required 206 to reach 0.8 power. For 2 participants, the consistency results were poor, ranging between 33% and 57%. However, for another two, 100% consistency was achieved. Minimum correctness resolutions were very small, ranging on average between 0.010 and 0.023 meaning that the models could successfully rank overall breast reconstruction outcomes that differed by 1.0-2.3%. However, maximum error resolutions were large, ranging on average from 0.172 to 0.305, meaning that the models incorrectly ranked outcomes that differed by 17.2-30.5%. Depending on the choice of preference model, the consistency for any one particular individual could vary by up to 16.7% with an average of 7.6%. Indeed, we examined the number of times that a particular model performed the best (or tied) with a given individual. The risk-averse multiplicative model performed the best with 22 participants (Table 3). We found one factor that might influence consistency: distribution of preference weights for attributes. There was a significant positive correlation with regard to weight distribution (R = 0.382, P = 0.024). In other words, the more equally a participant valued all the attributes, the less agreement between the participant and the preference models. The Spearman rank correlation between education and consistency was not statistically significantly different from zero (R = −0.2956, P = 0.080). However, the power for the given sample size was only 0.5, which suggests that the relationship with education may be need to be reassessed in the future with a sample size of at least N = 67. Time needed to complete the consultation did not seem to have an effect on agreement between the participant and preference models (R = −0.10, P = 0.57, power = 0.15, required N = 614). DISCUSSION Patients cannot be expected to evaluate every possible outcome in breast reconstruction. Through assessing and modeling their preferences with MAUT, we may computationally perform this vital step in breast reconstruction decision making. Previous studies have made untested assumptions when applying MAUT in the healthcare setting. We have found that these assumptions, while not incorrect, are too stringent in making decisions about breast reconstruction. Although most people tend to be risk-averse, the results demonstrate incorporating risk attitude is not critical in breast reconstruction preference modeling. Specifically, individual patients may be assumed to be risk neutral, which eliminates the need for risk attitude assessment, streamlines the consultation, and saves time. Contrary to previous assertions, 5,6,11 choice of a specific multiattribute utility function did not contribute to a significant difference in consistency. Furthermore, as the power additive function requires an additional risk tolerance assessment and The best-performing model was the multiplicative with risk-averse modeling. According to the Wilcoxon matched-pairs rank-sum test, the other models did not perform significantly worse than this index model. The number of times a model performed best for a given patient is also shown (best hits). NA, not available. Fig. 5. Boxplots of preference model consistencies. yields no improved consistency, it can be eliminated as a candidate. It would be highly unlikely to achieve 100% consistency for every participant. To begin, there is no perfect method for the elicitation of patient preferences or modeling these preferences. More importantly, preferences change: over the course of the consultation, several participants mentioned how their preferences were altered as they were forced to think about trade-offs (eg, they began to value out-of-pocket cost less and well-being more). It may be better to consider one's preferences over several days, 26,27,28 which is possible given that patients can have 2 or more encounters with their surgeon before surgery. In addition, several participants expressed appreciation that this process helped them understand their preferences. An interesting matter is the considerably high error resolutions that were observed. We assumed that the participants' rankings of outcomes were without error. However, they might misrank outcomes, by accident or embarrassment, resulting in abnormally large error resolutions. Upon examining several participants' preferences and rankings, we found several instances, which strongly indicated that this was the case. Taking into account the participant's preferences, we found that the participant's rankings subjectively did not make sense when they disagreed with the preference models at differences between outcome u-values above 0.10 (eg, reconstruction A was chosen by the participant over reconstruction B, even though B had a u-value 0.10 higher than A). If we assume that participants incorrectly ordered outcomes with error resolutions >0.10, then consistency improves another 14.8% on average (15.7% for the risk-averse multiplicative model). Therefore, the theoretical consistency may be as high as 94.6%. There are a number of sources that contribute to error in consistency (Fig. 6). We are only able to control for error originating from the preference models (hence this study) and, to a limited extent, preference elicitation. The correlation between consistency and preference weight distribution is not surprising. If a decision maker considers fewer attributes, making trade-offs becomes easier. Another way to view this is that the more strongly one feels about certain attributes, the easier it is to make trade-offs than if one were completely indifferent. No statistically significant correlation was found between consistency and education. However, the observed P-value was small (P = 0.080) despite the limited sample size (power = 0.57). There may be a negative relationship (R = −0.30) between years of education and consistency. It may be valuable to investigate this with a larger sample size; in particular, with more observations from the extremes (only high school education, doctoral degree). Although this type of analysis may be useful for most women preparing to undergo breast reconstruction, there is a small minority for whom these efforts would be unproductive given that MAUT was inconsistent with roughly 6% of the participants. CONCLUSIONS Breast oncologists and reconstructive surgeons spend a great deal of time and effort counseling Fig. 6. We attempted to evaluate consistency between the participant (patient model) and multiattribute utility theory (MaUt model). Both models begin with the same initial conditions (true participant subconscious preferences). However, the patient model has a feedback mechanism that we cannot model (eg, we have no direct access to a patient's subconscious or conscious preference). each block is a potential source of error. We only have control over "MaUt" and "consistency evaluation" and partial control over "preference elicitation. " note that reductions in consistency may be due to changes in preferences in the patient model due to feedback. Misranking outcomes is a major source of error in "consistency questions" on both the patient and MaUt side. it is difficult to discern if a participant indeed makes a mistake.	2016-05-12T22:15:10.714Z	2014-03-01T00:00:00.000	{ "year": 2014, "sha1": "8e0aa3ec3abbb8efecca88db259dbb94ff9960a3", "oa_license": "CCBYNCND", "oa_url": "https://doi.org/10.1097/gox.0000000000000062", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "8e0aa3ec3abbb8efecca88db259dbb94ff9960a3", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
4637235	pes2o/s2orc	v3-fos-license	Rapid identification of mycobacteria from positive MGIT broths of primary cultures by MALDI-TOF mass spectrometry Background Rapid identification of mycobacteria is important for timely treatment and the implementation of public health measures. The MGIT system ensures rapid detection of mycobacteria, but identification is usually delayed by days to weeks due to further subculture on solid medium. Matrix-assisted laser desorption time-of-flight mass spectrometry (MALDI-TOF MS) was demonstrated to effectively identify mycobacteria isolates subcultured from solid or liquid media. Reports of identification directly from MGIT broths of both sterile and non-sterile clinical specimens, omitting the subculture step, were limited and not satisfactory before. Our identification method dramatically shortened delay from detection to identification of mycobacteria. Methodology We assessed the performance of the Vitek MS IVD version 3.0 for direct identification of NTM and M.tuberculosis from primary MGIT cultures, and assessed two sample preparation methods. Results Direct identification of NTM from positive MGIT broths, using MALDI-TOF VITEK MS with IVD v.3.0, generated high rates of acceptable results reaching 96.4% (80/83), and up to 100% (83/83) for sample preparations including a 0.1% SDS washing step. The sensitivity of VITEK MS to identify M.tuberculosis from MGIT tubes was 58/72 (80.6%), when using immunochromatography (ICA) test as gold standard. A characteristic colony clumping, wool-like appearance was observed in 48, and all 58 (100%) were correctly identified as M.tuberculosis using MALDI-TOF. The detection rate of M.tuberculosis complex was low (10/24, 41.6%) in the 24 MGIT tubes that was polymicrobial. Our method significantly reduced both the reagent cost and turnaround time. Conclusions Based on a simplified protocol, we showed that MALDI-TOF MS can be used for rapid identification of NTM directly from primary MGIT cultures within the routine clinical laboratory workflow. However, we recommend an initial ICA test to screen for M.tuberculosis complex, due to a low identification rate of M. tuberculosis in the presence of polymicrobial cultures using MALDI-TOF. Results Direct identification of NTM from positive MGIT broths, using MALDI-TOF VITEK MS with IVD v.3.0, generated high rates of acceptable results reaching 96.4% (80/83), and up to 100% (83/83) for sample preparations including a 0.1% SDS washing step. The sensitivity of VITEK MS to identify M.tuberculosis from MGIT tubes was 58/72 (80.6%), when using immunochromatography (ICA) test as gold standard. A characteristic colony clumping, wool-like appearance was observed in 48, and all 58 (100%) were correctly identified as M. tuberculosis using MALDI-TOF. The detection rate of M.tuberculosis complex was low (10/ 24, 41.6%) in the 24 MGIT tubes that was polymicrobial. Our method significantly reduced both the reagent cost and turnaround time. PLOS time, however, these also require a high level of expertise and limited species can be identified. INNO-LiPA Mycobacteria (Innogenetics, Ghent, Belgium), targeting the 16S−23S rDNA spacer region, requires expensive equipment [23,24], but simultaneous detect and identify the genus Mycobacterium and 16 different mycobacterial species. GenoType Mycobacterium CM/ AS (Hain Lifescience, Nehren, Germany), targeting the 23S rDNA region, allows the detection of 31 species of NTM [25][26][27]. In recent years, Matrix-Assisted Laser Desorption Ionization Time-of-Flight (MALDI--TOF) mass spectrometry was demonstrated to accurately identify bacteria routinely isolated in a clinical microbiology laboratory [28][29][30]. The speed, robustness and minimal costs of sample preparation and measurement makes it exceptionally well suited for routine and highthroughput use. It reduces turnaround time and may potentially impact on benefiting patients. Numerous reports have described the performance of the Biotyper (Bruker Daltonics, Germany), Vitek MS RUO (formerly Saramis) and Vitek MS (bioMérieux, France) systems. Both systems use different algorithms for the identification of microbial protein spectra and have been shown to perform similarly for NTM identification [31,32]. In addition, with regular database expansion, the identification efficiency can be further enhanced. The latest released MBT Mycobacteria RUO Library version 5.0 on the Bruker Daltonic platform covers 164 of the currently known 180 mycobacteria species. However, the database is not FDA approved. VITEK MS contains a FDA-approved database IVD v.3.0 that includes 45 Mycobacterium species. In both databases, species belonging Mycobacterium tuberculosis complex (MTBC) allows only complex-level identification. For routine use, user friendliness is an important consideration in a clinical laboratory. The availability of disposable targets and ready-to-use matrix solution of the Vitek MS system reduce pre-analytical steps and possible errors. In addition, Vitek MS is easier to integrate into the workflow, using a common middleware (Myla™, bio-Mérieux) with other routine devices in our lab (Vitek 2, bioMérieux). Previous studies showed that reliable identification to species can be done using mycobacteria obtained from subculture on solid-phase medium, such as the Lowenstein-Jensen or Middlebrook 7H10 or 7H11 medium [31][32][33][34][35][36][37][38]. Identification rates using a Biotyper system was 77% [39] and detection rate increased from 67% to 94% by using the VITEK MS system for isolates grown on liquid media MGIT, using Saramis v4.12 and IVD v3.0 [40]. However, these studies all used mycobacterial isolates subcultured from either solid or liquid medium. In contrast, our study used mycobacteria isolates from MGIT broths, omitting the subculture step in identification. There are very limited number of studies reporting identification using isolates from MGIT medium without subculture, and showed that identification rates of NTM were poor when using a liquid medium [39][40][41]. Currently, there is only one published report using MALDI-TOF MS to identify NTM from 53 newly positive, liquid cultures of respiratory samples, and demonstrated a low rate of correct identification rate of 22% [42]. Identification of mycobacteria directly from MGIT broths can dramatically shorten the delay between detection by culture positivity to definitive identification of the mycobacteria. We therefore aimed to assess the performance of MALDI-TOF MS analysis for direct identification of Mycobacterium spp. from positive MGIT broths without subculture, compared to the routine protocol (with subculture), under real-world, routine laboratory settings. Materials and methods The study protocol was approved by the Institutional Review Board of Kaohsiung Veterans General Hospital, (No. 17-CT9-04). Samples We prospectively analyzed MALDI-TOF identification of mycobacteria directly from culture positive MGIT broths compared with the routine protocol, from Dec. 2016 to May 2017. All samples submitted for mycobacterial cultures were inoculated into the MGIT broth and cultured with MGIT Bactec MGIT 960 instrument (Becton Dickenson Cockeysville, Maryland, US) as well as Lowenstein-Jensen medium and incubated at 35˚C. Tubes flagged positive by the MGIT 960 instrument were removed from the instrument, subjected to the acid fast stain to evaluate the growth appearance as purity check and ICA (SD Bioline Ag MPT64 Rapid assay) test for rapid detection of M. tuberculosis complex. Positive MGIT broths that are negative for M.tuberculosis antigen using ICA were kept in room temperature for further identification to the NTM species at a monthly interval if considered clinically relevant [43]. Identification of MTB and NTM to species level by VITEK MS from the culture positive MGIT broth was compared to the results obtained by ICA test and Microchip array (Dr Chip Biotech Inc., Miao-Li, Taiwan), respectively, in the 6-months study period. All three methods were performed using the same liquid culture. Sample preparation for MALDI-TOF Vitek MS For identification of NTM, 1 ml of culture broth was used for both sample preparation methods (the direct method and the SDS method). To ensure sufficient amount of bacteria in the samples, the test was repeated using up to 3mL of the same MGIT culture broths, if the first try using 1mL failed. For identification of M. tuberculosis, 3 ml was used for both methods. The positive MGIT culture broths were vortexed for 5-10 seconds and 1 ml broth was transferred to a 1.5 ml Eppendorf tube. The Eppendorf tube was centrifuged at 14,000 rpm for 3 minutes, then the supernatant was completely removed. The pellet was either (a) subjected to ethanol inactivation without any washing step (the direct method), or (b) washed with 500 μL 0.1% Sodium Dodecyl Sulfate, centrifuged again at 14,000 rpm for 3 min before ethanol inactivation (the SDS method). The bacterial pellet was re-suspended with 500 μL of 70% ethanol and transferred to another 1.5 ml Eppendorf tube with 200 μL 425-600 um glass beads (SIGMA). After vortexing for 15 min and keeping in room temperature for 10 more minutes, the bacterial suspension was transferred to another 1.5 mL Eppendorf tube, centrifuged for 3 min to completely remove the supernatant. The pellet was mixed with 10 μL 70% formic acid by vortex for 3-5 seconds, then 10 μL 100% acetonitrile was added and mixed again by vortex for 3-5 seconds. The suspension was centrifuged at 14,000 rpm for 2 minutes and moved out of biosafety level III facility. One μL of the supernatant was spotted onto a MALDI-TOF target plate and air dried. MALDI-TOF MS Each deposit on the target plate was overlaid with 1 μL of matrix (VITEK MS CHCA) and air dried. The slide was run in the MALDI-TOF instrument (bioMérieux VITEK MS) to obtain the identification. All organisms were placed onto only one well of a Vitek MS slide. Target plates were calibrated and quality controlled both before and after data acquisition by using Escherichia coli ATCC 8739. After the acquisition of spectra, data were transferred from the Vitek MS acquisition station to the Vitek MS analysis server, and identification results were displayed using Myla v2.4 middleware. Each operator participating in the study was required to analyze a proficiency panel successfully prior to beginning to test isolates for this investigation. Microchip array MTB assay Microchip array MTB assay was performed according to the manufacturer's directions. Briefly, 0.5 ml of MGIT broth was washed and sonicated at 50˚C for 30 minutes. The vial was heated in boiling water for 20 minutes and then chilled on ice for 5 more minutes. After removal of cell debris, 5 ul aliquots of the extracted DNA were added to 20 μL of master mix and placed in a thermal cycler to perform the following program: 95˚C for 4 min (1 cycle); 95˚C for 20 s, 55 C for 20 s, and 72˚C for 40 s (15 cycles) and 95˚C for 20 s, 60˚C for 20 s, and 72˚C for 40 s (25 cycles); extension at 72˚C for 5 min. Mix 90 μL DR. HybTM Buffer and 10 μL amplicons in a new 1.5 mL Eppendorf vial. The vial was heated in boiling water for 5 minutes and then chilled on ice for 5 minutes immediately. The hybridization mixture was transferred into each well and covered with the plastic membrane. The plate was incubated at 55˚Cwith vibration for 1 hour in DR. Hyb™ oven (DR. Chip). The plastic membrane was removed, then placed on DR. Fluidic Station and execute "TB-SC" program under DR. Fluido software. The patterns were read and analyzed with DR. AiM Reader (DR. Chip). Discordant results Samples with discordant results between MS system and Microchip MTBC assay were further identified by polymerase chain reaction-restriction enzyme analysis (PRA). BstEII and HaeIII enzyme digestion of the amplification product was performed as described previously [44] and compared with PRA profiles published [10,[45][46][47]. Cost estimates We evaluated the cost per isolate for the MALDI-TOF mass spectrometry identification by adding the costs of Matrix reagents, target plates, and positive controls. All cost estimates are in U.S. dollars and reflect the actual costs incurred in our laboratory and not charges to the patient. respectively. (Table 1). MGIT broths that had MALDI-TOF results showing "bad spectrum" or "no identification", had a smaller pellet size after centrifugation than those with correct identification. Results In the routine workflow in our laboratory, positive MGIT tubes that were negative for M. tuberculosis using the ICA test, were kept in room temperature. NTM identification was performed monthly upon request or if the clinical criteria for NTM disease is met. As a result, MALDI-TOF MS was performed using cultures of different ages. Table 2 shows the rate of correct identification and the no of samples with different culture ages when identification was done, defined as the number of days between first detection by a positive MGIT culture to when identification testing was done. Although the age of the culture may influence the MAL-DI-TOF MS spectra, this delay did not affect the species-specific profiles. All the rapidly growing mycobacteria (n = 29) were correctly identified, with no association with the duration of culture positivity. However, in slow growing mycobacteria, a higher identification rate was associated with an increase in the number of days of culture positivity. Identification rate was increased to 96.4% (80/83) and 100% (83/83) for direct and SDS methods, respectively, when a larger sample size was used (up to 3mL of the same MGIT broth), producing a bigger pellet size of more than 2 mm along the side of Eppendorf tube. Notably, Microchip MTBC assay was unable to discriminate among the MAC complex. In contrast, MALDI-TOF VITEK MS was able to identify the 29 MAC strains to be 2 M. avium and 38 M. intracelluare. Species No Six MGIT broths that initially identified as M. genavense were confirmed to be M. intracellulare (1) and MTB (5) . Fig 2 shows that M. genavense has very few discriminating peaks, and resembled the peak profiles of some M.tuberculosis and M. intracellulare. Cost and turnaround time The reagent cost-per-bottle for direct identification from MGIT broth by VITEK MS compared to by microchip assay was reduced from US$12.4 to US $3.6. The turnaround time was reduced approximately from 4 hours to 60-70 minutes. Discussion Our study demonstrated that direct identification of NTM from a positive MGIT culture broth without subculture using VITEK MS was comparable to the Microchip MTB assay. VITEK MS correctly identified 96.4% of 83 positive MGIT broths with single isolates, which increased to 100% after treatment with 0.1% SDS; and Microchip MTB assay correctly identified 97.5%. One recent report which used MALDI-TOF MS to identify NTM from 53 newly positive liquid cultures of respiratory samples that were all monomicrobial, demonstrated a low rate of correct identification rate of 22% [42]. In that report, extraction started within 24 to 48 hours after the MGIT tube was flagged positive and only 1.2 mL of culture fluid was used for direct MALDI-TOF MS analysis. Only 7 (13.2%) were rapidly growing mycobacteria. In our study, rapidly growing mycobacteria were all identified successfully in the first week. Slowly growing mycobacteria had correct identification rate of 66.7% (6/9), and the rate increased with the duration that the MGIT broths were kept in room temperature. This implied that failure to identify mycobacteria using MALDI-TOF may be due to insufficient amount of bacteria in the sample. After increasing the volume of the culture fluid to 3 mL, high rates of identification rates were obtained. This was further strengthened by the protocol released recently by Biomerieux which suggests the use 3 mL of culture fluids for analysis. In addition, we also demonstrated that the identification rate can be further increased to 100% by washing the bacteria pellet using 0.1% SDS. Therefore, we suggest that the optimal sample preparation method is to use 3 mL of culture fluid and sample treatment with 0.1% SDS. The overall identification rate of M. tuberculosis from MGIT tubes by VITEK MS was lower (80.6%) than ICA. In pure cultures of M. tuberculosis, where typical cotton wool-like macroscopic appearance was observed in MGIT tubes, correct identification reached 100%. A low rate of identification (41.7%) was achieved when MGIT tubes were polymicrobial, where isolation of M. tuberculosis was mixed with NTM or other bacteria on subculture. A known limitation of MALDI-TOF MS is its inability to identify individual components in a polymicrobial culture. In countries where positive NTM cultures is frequent, there is a high possibility of isolating NTM in addition to M.tuberculosis complex in TB patients, which may lead to incorrect results with masking of TB. The reported incidence of positive NTM cultures was between 14.1-20.3 per 100,000 persons in year 2000-2003 in urban areas, such as Taipei [4], Ontario [49], and New York city [2]. The recovery of more than one mycobacteria from MGIT broths were not uncommon. Therefore, we suggest that MALDI-TOF analysis be used only for NTM identification after exclusion of M. tuberculosis by a negative ICA. M. genavense is a newly described pathogen with high levels of relatedness with M. triplex [50]. It causes disseminated infections in patients with AIDS. The clinical features mimicked those of disseminated M. avium complex infection, with invasion of liver, spleen and lymph nodes with acid-fast bacilli (AFB). [51] However, in our study, the initial identification of 6 strains of M. genavense was reconfirmed to be M. intracellulare (5) and M. tuberculosis complex (1). The misidentification may due to a low number of bacterial cells producing few signature molecules, resulting in peak profiles that mimics some M.tuberculosis and M. intracellulare, since M. genavense has very few discriminating peaks. We therefore suggest that identification of M. genavense should always be reconfirmed. The extremely high speed and low marginal cost of MS may improve laboratory efficiency if used directly on new positive liquid cultures. Identification by MALDI-TOF VITEK MS can be completed in approximately 1 hour after the MGIT broths were detected to be positive. Although it is not possible to perform sample preparation for MS each time mycobacteria is detected positive by the MGIT960 system without full automation of the procedure, it is feasible to perform the procedure once per day since the protocol can fit easily into the clinical laboratory workflow. NTM is ubiquitous in the natural and healthcare environment, and the clinical significance of isolating NTM in clinical specimens is difficult to determine. Rapid and accurate identification to the species-level may aid in guiding management decisions based on the pathogenic potential of the isolated species. Direct identification of NTM from positive MGIT culture broths can significantly shorten the turnaround time. The efficiency of mass spectrometry to identify mycobacteria grown on solid or liquid media was recently confirmed as a promising technique. VITEK MS v3.0 platform is an IVD system validated by manufacturer and provides many benefits. First, mycobacteria inactivated method recommended by the manufacturer only utilized simple materials, glass beads and 70% ethanol, to achieve safe, fast and effective inactivated results [52]. VITEK MS contains IVD-CE marked database for 49 species of mycobacteria, including 4 species in M. tuberculosis complex (Mycobacterium tuberculosis, Mycobacterium africanum, Mycobacterium bovis, and Mycobacterium canettii) and appeared relatively unaffected by the extraction method [33]. In this study, M. cometicum and M. mageritense were not included in the Microchip MTB database. Finally, we performed a cost analysis to determine whether the VITEK MS is economically competitive in the diagnostic laboratory. The reagent cost of MALDI-TOF was lower compared to the microchip assay, with a reduction from US$12.4 to US $3.6. There was a shortened turnover time from 4 hours using the microchip assay, to an average hands-on-time of 60-70 min per isolate for identification using MALDI-TOF [53,54]. Although the use of MALDI-TOF requires more expensive equipment and higher maintenance costs, the same system can be used for bacterial and yeast identification, which can reduce overall cost. In addition, the cost for identification instruments, supplementary biochemical tests and quality control testing for bacteria, yeast and mycobacteria can be reduced. It is important to take into consideration that the earlier identification of mycobacteria infections can result in other potential cost-savings, such as shorter hospital stays or better patient outcomes. Furthermore, VITEK MS runs user-friendly platform such as ready-to-use reagents, simplicity of operation [55]. These benefits may improve the workflow and turnaround time of mycobacterial identification and provide comprehensive operational requirement of clinical microbiology laboratory. The limitation of this study was that only patients that were considered potentially clinically relevant were included into the study for further identification of the positive MGIT broths. Mycobacteria species that were considered to be environmental mycobacteria with low pathogenic potential were not studied. However, this is the group of patients that are clinically important and requires identification of NTM to the species level. In conclusion, we demonstrated that the use of VITEK MS with IVD v3.0 is a reliable method for identifying NTM to the species level directly from culture positive MGIT broths without requiring subculture, which significantly shortened the turnaround time. We proposed an improved method for extraction of mycobacterial proteins using a bead-based disruption and the addition of 0.1% SDS washing step which achieved high rates of correct identification. However, we recommend that the detection of M. tuberculosis should be done initially by using ICA because it is easy to perform, can be done within the BSL-3 facility, and most importantly, its performance is not influenced by the presence of multiple bacteria in the specimen. We also showed that the most important factor for correct identification by MAL-DI-TOF is the amount of bacteria and not the freshness of the culture fluid. Our results provides promise for application and incorporation of MALDI-TOF MS into routine use in the clinical setting for rapid identification of NTM directly from culture positive liquid MGIT broths.	2018-04-03T01:28:11.875Z	2018-02-02T00:00:00.000	{ "year": 2018, "sha1": "f6638101853b3c7e7d7e9c72740dd82c6c60989b", "oa_license": "CCBY", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0192291&type=printable", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "f6638101853b3c7e7d7e9c72740dd82c6c60989b", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
5015244	pes2o/s2orc	v3-fos-license	Secondary Degeneration of Auditory Neurons after Topical Aminoglycoside Administration in a Gerbil Model Hair cells in the cochlea can be damaged by various causes. Damaged hair cells can lead to additional destruction of parts of the auditory afferent pathway sequentially, which is called secondary degeneration. Recently, researches regarding cochlear implants have been actively carried out for clinical purposes; secondary degeneration in animals is a much more practical model for identifying the prognosis of cochlear implants. However, an appropriate model for this research is not established yet. Thus, we developed a secondary degeneration model using an ototoxic drug. 35 gerbils were separated into four different groups and kanamycin was applied via various approaches. ABR was measured several times after drug administration. SGCs were also counted to identify any secondary degeneration. The results showed that outer and inner HCs were damaged in all kanamycin-treated groups. Twelve weeks after kanamycin treatment, the round window membrane injection group showed severe subject differences in hair cells and SGC damage, whereas the gelfoam group showed consistent and severe damage in hair cells and SGCs. In this study, we successfully induced secondary degeneration in hair cells in a gerbil model. This model can be used for various purposes in the hearing research area either for treatment or for preservation. Introduction Social impacts of hearing loss have increased in many aspects more than ever, since the prevalence of hearing loss surges in accordance with the aging process of our modern society. Noise, ototoxic drugs, infections, aging, and other diseases are responsible for cochlear end organ damage during our lifetimes. In many cases, the loss of cochlear hair cells is the main contributor to loss of sound perception. Cochlear hair cell damage can subsequently progress towards the proximal part of the auditory pathway including the nerve fiber, spiral ganglion cells (SGCs), and cochlear nucleus, which is also known as secondary degeneration [1]. This secondary degeneration shows various features in terms of the degree and rate of degeneration depending on etiologies of hair cell damage and species [2][3][4][5][6]. Specifically, this degeneration had been considered to be highly dependent on the status of the inner hair cell (IHC) [2,7,8]. Supporting cells, which are located under the inner hair cell, were also considered as an important factor that can contribute to the degree and time of secondary degeneration in both animals and humans [9,10], and this was further supported by a study with transgenic mice [11]. Even after severe hearing loss, the degree of secondary degeneration on the remaining SGCs is very critical for hearing rehabilitation in the area of cochlear implant which is a cutting-edge modality for profound sensorineural hearing loss patients nowadays [12,13]. Currently, cochlear implants rely on SGCs for electrical stimulation for coding of the processed acoustic sound, which means a higher hearing performance can be expected with a higher number of SGCs [14]. This urges the clinical modality to prevent or retard the secondary degeneration of SGCs while waiting for a cochlear implant surgery. Aminoglycoside is a widely used class of antibiotics which also has ototoxicity that can induce permanent damage to the organ of Corti (OC) [15]. Particularly, kanamycin is more cochlear-toxic rather than vestibulotoxic [16] and has 2 BioMed Research International been used in animal research for deafening [17,18]. Several studies tracked the histological feature of the auditory afferent pathway after kanamycin deafening [19], especially when administered with furosemide [20] that also has a potential of causing hearing loss [21]. Kanamycin can be accumulated mainly in the mitochondria of HCs [22] which can result in SGC loss by affecting the neurotrophic factors [23][24][25]. As described above, delaying and attenuating the timing of secondary degeneration are important for hearing rehabilitation. To investigate the therapeutic methods which can delay or prevent secondary degeneration, a stable and consistent secondary degeneration model is essentially needed. For this reason, the purpose of this study is to establish a consistent secondary degeneration model by approaching different drug treatments of kanamycin. Drug Administration. Kanamycin (KM) sulphate was diluted in normal saline (150 mg/ml) and administered in three different ways ( Figure 1). For the KP group, 50 ul of the drug solution was delivered by injection with an insulin syringe (Ultrafine Insulin Syringe, Becton Dickinson, USA) to the bulla. To improve the absorption to the RWM, the animal was laid on its contralateral side, injected, and sustained for half an hour. For both KG and KI groups, the bulla was exposed through a retroauricular skin incision. After anesthetization, furs near the bulla were removed and the skin was also incised. A small hole was made on the bulla and the RWM was exposed. A small gelfoam was placed on the RWM and 4 microliters of kanamycin solution was injected with a Hamilton syringe (Hamilton Company, Nevada, USA) for the KG group. In the case of the KI group, a small hole was made on the bulla and the tip of a cannula connected with the Hamilton syringe was fitted into the RWM after the endolymph liquid was drained. Then, 4 microliters of the KM solution was gently and slowly injected. Animals in both groups were also laid on their contralateral side for surgery for stable absorption of the drug. Hearing Measurement. Auditory brainstem responses (ABRs) were measured to investigate the changes of hearing threshold before and after drug administration. The evoked response signal-processing system (System III; Rucker Davis Technologies, Alachua, Florida) was adopted for ABR measurement. Animals were anesthetized with zolazepam (Zoletil, Virbac, Carros Cedex, France) and xylazine (Rompun, Bayer, Leverkusen, Germany) and were placed in a soundproof chamber. Then, needle electrodes were inserted to the vertex (as response) and ventrolateral sides of both pinnae (as reference and ground). Tone stimuli with 4, 8, 12, 16, and 32 kHz were generated from 90 dB to 10 dB in 5 dB steps and average waveforms were generated from 1024 responses. Hearing thresholds were measured before and 1, 4, and 12 weeks after drug administration. Histological Analysis and Quantification. Animals were sacrificed after 4 and 12 weeks of drug administration for the histological analysis. Cochleae were harvested and fixed in 4% paraformaldehyde for 24 hours at 4 ∘ C. After fixation, the samples were decalcified with 0.5 M ethylenediaminetetraacetic acid (EDTA) for a week. Cochleae from 6 gerbils in each group were embedded in paraffin and sectioned from the apex to the basal turn to quantify the number of SGCs at 4 and 12 weeks after drug administration. Four micrometers of sectioned samples was stained with hematoxylin and eosin (H&E). The number of SGCs was counted using ImageJ software (http://rsb.info.nih.gov/ij/). Samples were sectioned as midmodiolar plan to represent the overall place of the cochlea. SGCs in 10000 square micrometers at four different parts of the cochlea (high middle, low middle, high basal, and low basal) were counted and compared with the control group. Three or more sectioned images at each part with a 50 um interval were averaged and examined by polarizing microscopy using a BX51-P microscope (Olympus, Tokyo, Japan). Three cochleae at each group after 12 weeks of drug administration were prepared as whole mounts and were immunostained with anti-neurofilament heavy (anti-chicken, Millipore, 1:1000) and MyosinVIIa (anti-rabbit, Millipore, 1:200). After mounting on a slide, images were taken with a confocal microscope (FV-3000, Olympus, Tokyo, Japan). 40x magnification was used and -stacks were generated. Statistical Analysis. Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS, Chicago, USA) software. Two-way analysis of variance (ANOVA) with Bonferroni post hoc test was adopted and significant differences were determined when the value was <.05. Serial Changes of ABR Thresholds. We tested different delivery techniques (single application) to induce hearing loss and evaluated the hearing outcome. ABR was measured at three time points in all groups, one, four, and twelve weeks after the drug administration. Hearing threshold shifts were observed in all groups at any given time points (Figure 2). At the one-week time point, hearing thresholds of all kanamycin injected groups were statistically different from the control (two-way ANOVA, df = 3, = 193.1, < .0001). In the KP group, post hoc -test revealed that statistical significance exists at 12, 16, and 32 kHz (12 kHz: < .001, 16 kHz: < .001, Figure 1: Drug delivery via different methods. The KM solution was injected by a syringe to the KP group (a). A small hole on the bulla was made and the RWM (black arrowhead) was exposed (b). A gelfoam with KM was placed on the RWM for the KG group (c). The tip of a cannula was inserted in the RWM niche for the KI group (d). One week a er drug administration In the KP group, hearing thresholds increased at high frequency regions (12, 16, and 32 kHz) at four weeks after drug administration and these increased thresholds remained until 12 weeks after drug administration. KG group showed a more severe hearing threshold change at all time points. KI group showed the worst hearing thresholds change at all tested frequencies and all test points ( * * * < 0.001). (Figure 2). According to the results of this part, using the local delivery techniques to induce a substantial hearing threshold shift with a single application was possible. Threshold changes of the KG and KP groups showed increased hearing threshold shifts at higher frequencies due to the closer location to the round window where the drug is presumably delivered. In the KI group, most of the deterioration in the hearing threshold was observed; however, the invasiveness of the delivery technique cannot be disregarded. At the 4-week time point, hearing thresholds of all kanamycin injected groups were statistically different from the control (two-way ANOVA, < 326.9, df < 3, < .0001). Post hoc -test revealed that all three groups showed With a single application using a different delivery technique, hearing threshold change was maintained until the 12week time point. This result suggests that the hearing deterioration observed at the 1-week time point is not transient but is permanent, possibly due to the irreversible loss and not the temporary damage of the hair cells. Hair Cells and Neurofilaments Damage Was Variable Depending on the Drug Administration Methods. A decrement in hearing threshold after drug administration would be highly related to the status of the OC, especially hair cells (HCs) and nerve fibers connected to them. Thus, we investigated the status of the HC and nerve fibers by immunostaining with whole mount preparation. The status of HCs and nerve fibers at the apex, middle, and base parts of the cochlea, which represents 8, 16, and 32 kHz, was identified. After twelve weeks of drug administration, the KG and KI groups showed a total loss of HCs and a partial loss of nerve fibers at three selected parts of the cochlea. In the case of the KP group, we found that the status of the nerve fiber and HCs was preserved (Figure 3). These results suggest that the HC and nerve fiber can be severely damaged by a single application of kanamycin depending on the delivery method, and such damage would cause a permanent threshold shift in the KG and KI groups. In the case of the KP group, a threshold shift was maintained for 12 weeks without anatomical change in the HCs and nerve fibers. OC Was Damaged over Time after Drug Administration. According to the immunostaining results, kanamycin causes damage not only to the HCs in the OC, but also to the auditory nerve fibers. To investigate the degree of degeneration in the OC and the possibility of additional degeneration of the auditory ascending pathway, the status of the OC within the sectioned images was identified. Four locations in the cochlea were selected as representative areas (Figures 4 and 5). After four weeks of drug administration, the OC was intact at the four selected locations in the control and KP groups. In the KG group, the OC was damaged and showed a flat epithelium at both high and low basal parts of the cochlea. In the case of the KI group, the OC was also damaged and showed a flat epithelium throughout the cochlea (Figure 4). This result confirmed that a single administration of kanamycin can cause damage to HCs at 4 weeks after the treatment depending on the administration method. After twelve weeks of drug administration, still, the status of the OC was intact at all locations in the control and KP groups. In the KG group, damage to the OC was extended to the upper parts of the cochlea, but the high middle part of the OC was undamaged. In the KI group, the status of the OC was varied depending on the subject. One subject showed a total loss of the OC at all the selected parts, whereas two subjects showed an intact status of the OC. These results suggest that the kanamycin solution injected through the RWM would have leaked out if the perilymph was not well flushed. at four locations were counted and compared with the control group (Figures 3 and 4). In the KP group, there were no difference in the density of the SGC at four and twelve weeks after drug administration ( Figure 6). However, the number of SGCs in the KG group was significantly decreased compared with the control group at four weeks after drug administration (Figure 6(a)), and these decrements were increased at twelve weeks after drug administration ( Figure 6(b)). In the KI group, the number of SGCs was significantly decreased in all selected locations at four weeks after drug administration (Figure 6(a)). However, after twelve weeks of drug administration, the number of SGCs was not consistent between subjects and a huge subject difference existed and was not significantly different from the control group ( Figure 6(b)). These results suggest that a single treatment of kanamycin can damage HCs and this HC loss causes deterioration at the upper part of the auditory pathway. KM Ototoxicity. Kanamycin is a well-known ototoxic agent, and it is a widely used model to mimic human sensorineural hearing loss with various delivery methods in an animal model [20,[26][27][28][29]. However, to induce substantial hearing loss in a rodent model, multiple injections or other drug combinations are required. Several previous researches used kanamycin as an ototoxic drug combined with other agents for achieving total loss of HC which is incomplete with kanamycin alone [5,30]. In particular, furosemide, which can manipulate the blood labyrinth barrier, has been used as a combination agent and showed better ototoxic damage compared to kanamycin alone. However, this agent is a diuretic and can systemically affect the whole body. This untargeted effect would cause changes inside the cochlea, resulting in ambiguous toxic or inflammatory damage. It was reported that furosemide may itself cause a hearing loss by inducing transient malfunction in the stria vascularis [31,32]. Therefore, we used a very high concentration of kanamycin and applied it directly to the round window membrane with a gelfoam and acquired severe SGN loss at 4 weeks after drug administration. This high concentration of kanamycin was two or three times higher than in previous studies [5,30] and causes total loss of both hair cells and SGCs at 12 weeks after a single treatment. The ototoxicity of kanamycin is a well-known issue throughout the clinical and animal research. It is reported that kanamycin induces production of reactive oxygen species and these attack the cochlear hair cells, which is an irreversible injury, resulting in a hearing disorder (Jiang et al., 2005). We did not explore the mechanism of kanamycin toxicity at the HC and SGC in this study. Nevertheless, we might expect that the same ototoxic mechanism would be involved in this study. Gerbil as a Proper Hearing Research Model. The Mongolian gerbil is a well-established animal model for hearing research [33]. Since the range of audible frequency is more similar to that of humans than other rodents such as mice, rats, or guinea pigs [34], gerbils have been considered as a suitable model in hearing research including aminoglycoside toxicity [5]. Furthermore, the larger bulla and thin skull enable a surgical approach to the cochlea. In our study, we applied various approaches to treat with aminoglycosides and acquired consistent results after surgery with the aforementioned reasons except with the RWM injection group. Additionally, due to the location of the stapedial artery that is not blocking the round window, gerbils have been widely used as an animal model for CI studies. Moreover, the high reproduction rate and easier breeding and handling make gerbils more appropriate for animal studies. Their characteristics (i.e., they were born deaf and have a late onset of hearing [35]) boost the versatility for various approaches that are possible with this animal model. All in all, we suggest 8 BioMed Research International that gerbils can serve as a proper animal model in various hearing research areas. Drug Delivery Agent. We applied various ways of drug administration to find the most appropriate way that can induce secondary degeneration. According to our results, the RWM injection showed a more dramatic change in the OC within a short time than the gelfoam group. The number of SGCs was severely decreased at four selected parts of the cochlea at 4 weeks after kanamycin treatment through the RWM injection. However, a prominent subject difference also existed with this approach. During treatment, the KM solution could not disperse well because of the pressure inside the scala tympani, and it also leaked out right after injection using a cannula. Together with these, we considered that KM RWM injection is not an appropriate way to create a secondary degeneration model ( Figure 6). Otherwise, KM application with gelfoam showed very consistent and effective results within subjects compared to any of the other methods ( Figure 6). Drug delivery agents have been studied in the otology research area for ototoxic or therapeutic purposes. It has been reported that gelfoam increases the effect of a drug itself by allowing the drug to be retained longer at the target area. Abbas and Rivolta (2015) used aminoglycosides with a gelatin sponge and reported a more significant change in the hearing threshold after 2 weeks of treatment than with KM alone [5]. However, when they applied gentamycin to the RWM with a gelatin sponge, which is also a well-known ototoxic drug, it did not cause hearing loss. They explained that this is due to the polar nature of the gelatin sponge which blocked the penetration of gentamycin into the RWM [5]. Poloxamer 407 has also been used as a delivery agent, including nanoparticles and an ototoxic drug, which can provide sustained release at the target area [36,37]. Together with these, we thought that kanamycin treatment with poloxamer 407 would also be a proper way to create a secondary degeneration model in gerbils. Secondary Degeneration Modeling and Possible Treatment. For generating a secondary degeneration model for diverse purposes, various methods would be applicable in animal research. A recent study that investigated KM toxicity reported that KM did not damage adult spiral ganglion neurons [4]. In this study, hair cells and ganglion neurons in postnatal day 3 rat cochlear organ culture were damaged by KM, whereas there was no toxicity in adult rat ganglion cells in an organotypic culture model. This result supports the notion that the degeneration of SGCs in our study was damaged not by KM itself but influenced from the loss of HC as secondary damage. It also supports the fact that our reproducible secondary degeneration model with the gelfoam approach is suitable for a secondary degeneration model. As seen in Figure 3, the KP group showed an intact histological structure of HCs and neurofilaments, even with a hearing threshold decrement 12 weeks after drug administration. This result suggests the possibility that secondary degeneration could be initiated at the upper level of the HC. It was reported that a low dose of aminoglycoside causes synaptic changes without HC loss [38]. If the secondary degeneration starts from the damage synapse level which is the most vulnerable factor in the cochlea, neurotrophin and photobiomodulation would be appropriate therapeutic approaches. Neurotrophin factor, especially NT-3, was reported as a very useful agent which can protect against synaptic loss due to noise exposure [36,39]. Photobiomodulation with a lowlevel laser (LLL) has also been studied in hearing research areas, and it was reported that it has a protective effect against HC after noise exposure [40,41]. Furthermore, it has a neuroprotective effect against Ouabain on SGC [42]. We will try to protect or delay the secondary degeneration with either neurotrophin or photobiomodulation in the near future. Conclusion We induced secondary degeneration of HCs in a gerbil model through diverse drug delivery approaches in this study. High concentrations of kanamycin application with gelfoam on the RWM caused severe HC loss and this extended to degeneration of the auditory nerve and SGCs. This model can be used for various purposes in the hearing research area either for treatment or for preservation. Furthermore, this model would be applicable for research regarding cochlear implants.	2018-05-03T00:19:19.814Z	2018-03-01T00:00:00.000	{ "year": 2018, "sha1": "478d2650180a8b3cbb01dd064eb9573fec99a42c", "oa_license": "CCBY", "oa_url": "http://downloads.hindawi.com/journals/bmri/2018/9158187.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "50524b5139670a4c7d152dec2c373fdbec4a682e", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
231979427	pes2o/s2orc	v3-fos-license	Generative Speech Coding with Predictive Variance Regularization The recent emergence of machine-learning based generative models for speech suggests a significant reduction in bit rate for speech codecs is possible. However, the performance of generative models deteriorates significantly with the distortions present in real-world input signals. We argue that this deterioration is due to the sensitivity of the maximum likelihood criterion to outliers and the ineffectiveness of modeling a sum of independent signals with a single autoregressive model. We introduce predictive-variance regularization to reduce the sensitivity to outliers, resulting in a significant increase in performance. We show that noise reduction to remove unwanted signals can significantly increase performance. We provide extensive subjective performance evaluations that show that our system based on generative modeling provides state-of-the-art coding performance at 3 kb/s for real-world speech signals at reasonable computational complexity. INTRODUCTION In recent years it has become possible to generate high-quality speech from a conditioning sequence with a very low information rate. This suggests that generative synthesis forms a natural basis for the coding and enhancement of speech. However, it has been found that generative synthesis is sensitive to the quality of the data used for training and the conditioning sequences used for training and inference. This can result in poor synthesized speech quality. In this paper, we discuss methods that significantly reduce the impact of distortions in the input signal on signal synthesis for speech coding. WaveNet [1] first showed that the generation of high-quality speech from only a low-rate conditioning sequence, such as written text, is possible. WaveNet is based on an autoregressive structure that specifies a predictive distribution for each subsequent signal sample. While WaveNet uses a dilated convolution to determine the predictive distribution, other recurrent neural networks structures such as WaveRNN [2] and the WaveGRU structure that we use in this paper have also been used successfully for this purpose. Although autoregressive structures for synthesis are common, feedforward structures are used by, for example, Parallel WaveNet [3], WaveGlow [4], WaveGAN [5], and GANSynth [6]. It is fair to state that while the more recent methods may have computational advantages, they do not surpass the basic synthesis quality of the original WaveNet approach. The high quality of generative speech synthesis has led to a significant effort towards its usage for coding. In contrast to synthesis from text, synthesis for coding must be able to generate an unlimited range of voices and its conditioning is variable as it is computed from input signals that may suffer from a range of distortions. It was found that the synthesis of a wide range of voices with a single generative model is not a significant problem. Generative synthesis of a wide range of unknown voices with a single model results in only a minor reduction of speaker identifiability [7]. However, variability in the conditioning leads to a reduced output speech quality that can not be improved significantly with straightforward measures such as training with noisy conditioning and undistorted target signals. Hence, despite extensive research, e.g., [8][9][10][11][12] generative synthesis based speech coding has not yet seen major practical applications. The contribution of this paper consists of the identification of causes of the sensitivity to distortion, the development of methods to reduce this sensitivity, and subjective testing of the new methods confirming the improvements. We show that a major cause of the sensitivity is associated with an attribute of the log-likelihood (LL) objective function. The LL objective function incurs a high penalty if the model assigns a low probability to observed data. Hence, in the context of autoregressive structures, it encourages an overly broad predictive distribution when at least some training data are difficult to predict accurately from the past signal and conditioning, which is true for real-world training data. We mitigate this effect by including predictive variance regularization in the objective function. We also show with experiments that input-noise suppression can improve performance significantly. It is well-known that a sum of low-order linear autoregressive processes is, in general, not a loworder autoregressive process. This suggests that linear autoregressive models are poor for sums of independent signals and our results indicate this also holds for nonlinear autoregressive models. Whereas traditional analysis-by-synthesis coding methods can compensate for model inadequacies, this is not true for generative synthesis based coding, explaining the effectiveness of noise suppression. PROBLEM FORMULATION In this section, we first describe how an autoregressive model is used to model a process. The method is as proposed in [1]. We then discuss a common problem that occurs when training such sequences. Consider a random process {Xi} consisting of real-valued random samples Xi, with a time index i ∈ Z. The joint distribution of a finite sequence, p(xi, · · · , xi−N ), can be expressed as a product of conditional distributions: where β is conditioning information. It follows from (1) that we can create an approximate realization of a random process by recursively sampling from a model of the predictive distribution p(xi\|xi−1, · · · , xi−N , β) for sufficiently large N . It is convenient to use a standard-form distribution q(xi\|α) with parameters α as a model predictive distribution. The standardform distribution can be a Gaussian or a logistic mixture, for example. This formulation allows us to predict the model parameters with a deterministic neural network φ : (xi−1, · · · , xi−N , β, W ) → α where W is a vector of network parameters. Thus, the predictive distribution for sample xi is now q(xi\|φ(xi−1, · · · , xi−N , β, W )). To find the parameters W , a reasonable objective is to minimize the Kullback-Leibler divergence between the ground truth joint distribution p(xi, · · · , xi−N ) and the model distribution q(xi, · · · , xi−N ), or, equivalently, the cross-entropy between these distributions. The latter measure is tractable even though p is only available as an empirical distribution. It follows from (1) and our formulation of q(xi\|α) that cross-entropy based estimation of the parameters of φ can be implemented using maximum-likelihood based teacher forcing. For a database of M signal samples, the maximum-likelihood estimate of W can be written as Note that (2) leads to rapid training as it facilitates parallel implementation. For sufficiently large N and M , the LL objective provides an upper bound on the differential entropy rate as where, for notational convenience, we considered the unconditioned case. Conversely, (3) can be interpreted as a lower bound on a measure of uncertainty associated with the model predictive distribution. This lower bound is associated with the process itself and not with the model. Although the differential entropy rate is subadditive for summed signals, predictive models tend not to work well for summed signals. In general, a model of summed signals is essentially multiplicative in the required model configurations. It is well-known that the sum of finite-order linear autoregressive models is, in general, not a finiteorder autoregressive model [13]. It is relatively straightforward to reduce this problem with noise suppression. A more difficult problem relates to well-known drawbacks of the Kullback-Leibler divergence and, hence, the LL objective of (2). When the model distribution q vanishes in the support region of the groundtruth p, the Kullback-Leibler divergence diverges. In (2) this manifests itself as a severe penalty for training data xi that have a low model probability q(xi\|φ(xi−1, · · · , xi−N , β, W )). Hence, a few nonrepresentative outliers in the training data may lead the training procedure to equip the predictive model distribution with heavy tails. Such tails lead to signal synthesis with a relatively high entropy rate during inference. In audio synthesis this corresponds to a noisy synthesized signal. Hence it is desirable to counter the severity of the penalty for low probability training data. We can identify a second relevant drawback to the ML objective. When the ML objective function is used, the model distribution should converge to the groundtruth distribution with increasing database size. However, in practice the stochastic nature of the training data and the training method results in inaccuracies and this in turn means the method attempts to minimize the impact of such errors. For example, the implicit description of pitch by the predictive distribution may be inaccurate. A predictive model distribution with heavy tails for voiced speech then increases the likelihood of training data as it reduces the impact of the model pitch deviating from the groundtruth pitch. From this reasoning we conclude that it is desirable to account for the audibility (perception) of distortions, leading to empirically motivated refinements of the objective function. The problems associated with the LL objective have been considered earlier in different contexts. The vanishing support problem described above was addressed in the context of generative adversarial networks (GANs) [14], where the implicit Jensen-Shannon objective function of the original method and the more general fdivergence based method [15] suffer, at least in principle, from similar problems. The support problem in GANs can be removed by using the 1-Wasserstein distance [16] or with maximum mean discrepancy (MMD) [17,18]. However, as these measures require as input two empirical distributions, these methods are natural for static distributions and not for dynamic predictive distributions. The methods also do not facilitate adjustment to account for perception. An existing approach that attempts to compensate for overly broad predictive distributions is to lower the "temperature" during inference, e.g., [19]. The predictive distribution is typically raised to a power, then renormalized. This approach does not account for the implicit cost penalty in the basic training objective. OBJECTIVE FUNCTIONS FOR PREDICTIVE DISTRIBUTION MODELS In this section, we discuss two related approaches that modify the maximum likelihood criterion to obtain improved performance. Both approaches aim to reduce the impact of data points in the training set that are difficult to predict. The methods remove the need for heuristic modifications during inference. While the principles of our methods are general, we apply them to the mixture of logistics distribution that we use in our coding scheme (cf. section 4). Encouragement of low-variance predictive distributions We now discuss how to add a term to the objective function that encourages low-variance predictive distributions. In this approach we define the overall objective function for the weights W given a database {x} as where the log likelihood over the database, JLL({x}; W ) = E data log q(xi; φ(xi−1, · · · , xi−N , W )), is combined with a variance regulatization term Jvar({x}, W ) that is defined below and where ν is a constant that must be tuned. Predictive variance computation The variance of the predictive distribution is an instantaneous parameter that varies over a database and Jvar({x}, W ) must be an average over the predictive distributions. The predictive distribution of each sample has a distinct variance and the averaging method can be selected to have properties that are advantageous for the specific application. As noted in section 2, the predictive distribution is a standard-form distribution q(x\|α). The predictive distribution q(x\|α) is commonly a mixture distribution. Hence we must find an expression for the variance of a mixture distribution. We first note that the mean of a mixture distribution is simply where Eq is expectation over q and q k =q(·; µ k , s k ), withq a mixture component. The variance of the mixture distribution is We now consider the specific case of a mixture of logistics in more detail. The logistic distribution for component k is: where s k is the scale and µ k is an offset. It is easily seen that the logistic distribution is symmetric around µ and that, hence, µ is the distribution mean. The variance of the logistic distribution is We can now write down the variance of the mixture of logistics model by combining (6) and (8): Regularization terms The most obvious approach for reducing the prediction variance is to use the prediction variance (9) directly as variance regularization in the objective function (4): where E data indicates averaging over the database. That is, we encourage selection of weights W of the network φ that minimize σ 2 q . Straightforward optimization of (10) over a database may result in the prediction variance being reduced mainly for signal regions where the conditional differential entropy (3) is large. The conditional differential entropy can be decomposed into the sum of a scale-independent term and a logarithmic scale (signal variance) dependency. For speech the scale-independent term is large for unvoiced segments while the scale-dependent term is large for voiced speech (as it is relatively loud). For signals that have uniform overall signal variance, it may be desirable to encourage low predictive variance only for regions that have relatively low conditional differential entropy. (For speech that would correspond to encouraging low variance for voiced speech only.) This can be accomplished by a monotonically increasing concave function of the predictive variance. The logarithm is particularly attractive for this purpose as it is invariant with scale: the effect of a small variance getting smaller equals that of a large variance getting smaller by the same proportion. We then have: with a providing a floor. Baseline distribution approach For completeness we describe an alternative method for preventing the vanishing support problem of the Kullback-Leibler divergence by using a "baseline" distribution. To this purpose consider a mixture distribution of the form qtrain(xi; φ) = γ0q(xi\|α0)+ K k=1 γ kq (xi; φ(xi−1, · · · , xi−N , β, W k ), (12) where the parameters γ0 and α0 are set by the designer and where the first term is omitted during inference (the other terms must be renormalized by a factor 1 1−γ 0 ). By selecting α0 to provide an overly broad distribution, the distribution used for inference will be of low variance. SYSTEM ARCHITECTURE In this section we describe the architecture of our coding scheme. The parameter settings of the scheme are provided in section 5.1. Let us consider an input signal with a sampling rate S Hz. To avoid the need for modeling summed independent signals, the input is pre-processed with a real-time TasNet [20,21]. The encoder first converts the signal into a sequence of log mel spectra (e.g., [22]). A set of subsequent log mel-spectra are stacked into a supervector that is subjected to a Karhunen-Loève transform (KLT) that is optimized off-line. The transformed stacked log mel spectra are encoded using split vector quantization with a small number of coefficients per split. No other information is encoded. The decoder first decodes the bit stream into a sequence of quantized log mel spectra. These spectra form the input to the conditioning stack, which consists of a set of 1D convolutional layers, all except the first with dilation. The output is a vector sequence with a sampling rate equal to that of the mel spectra of the encoder and a dimensionality equal to the state of the GRU unit discussed below. The autoregressive network consists of a multi-band WaveGRU, which is based on gated recurring units (GRU) ) [23]. For our Nband WaveGRU, N samples are generated simultaneously at an update rate of S/N Hz, one sample for each frequency band. For each update, the state of the GRU network is projected onto an N ×K ×3 dimensional space that defines N parameter sets, each set corresponding to a mixture of logistics for a band. The value of a next signal sample for each band is then drawn by first selecting the mixture component (a logistics distribution) according to its probability and then drawing the sample from this logistic distribution by transforming a sample from a uniform distribution. For each set of N samples a synthesis filter-bank produces N subsequent time-domain samples, which results in an output with sampling rate S Hz. The input to the WaveGRU consists of the addition of an autoregressive and conditioning components. The autoregressive component is a projection of the last N frequency-band samples projected onto a vector of the dimensionality of the WaveGRU state. The second component is the output of the conditioning stack (dimensionality of the WaveGRU state), repeated in time to obtain the correct sampling rate of S/N Hz. The training of the GRU network and the conditioning stack is performed simultaneously using teacher forcing. That is, the past signal samples that are provided as input to the GRU are groundtruth signal samples. The objective function ) , combining log likelihood (cross entropy) and variance regularization, is used for each subsequent signal sample. For our implementation with variance regularization, we found the baseline distribution not to aid performance significantly, and it was omitted from the experiments. EXPERIMENTS Our experiments had two goals. The first is to show the effect of predictive variance regularization and noise suppression. The second is to show that our contributions enable a practical system. System configuration We tested eight systems, all variants based on a single baseline system operating on 16 kHz sampled signals. It is conditioned using a sequence of 160-dimensional log mel spectra computed from 80 ms windows at an update rate of 50 Hz. The system uses four frequency bands, each band sampled at 4 kHz. The conditioning stack consists of a single non-causal input layer (expanding from 160 channels to 512 channels), three dilated causal convolutional layers with kernel size two, and three upsampling transpose convolutional layers (kernel size two). The overall algorithmic delay is 90 ms. The conditioning outputs are tiled to match the GRU update frequency. The GRU state dimensionality is 1024, and eight mixture-of-logistics components are used for the predictive distribution per band. The systems were trained from randomly initialized weights W for 7.5 million steps, using a mini-batch size of 256. The target signal was from a combination of clean [24,25] and noisy sources [26], including large proprietary TTS datasets. Additional noise was added from [27], with random SNR between 0 and 40 dB SNR. Table 1 shows the combinations of coder attributes that were used. We briefly discuss each attribute. The variance regularization included refinements that further improved its performance: it was applied to the first two bands only and ν in (4) was made proportional to a voicing score. The noise suppression system was a version of ConvTasNet [21]. The weight pruning attribute was selected to enable implementation on consumer devices. For the three GRU matrices, we used block-diagonal matrices with 16 blocks, which uses 93% fewer weights than a fully connected model. For other hidden layers, we applied iterative magnitude pruning to remove 92% of the model weights [28]. The pruning makes the codec with TasNet run reliably on a Pixel 2 phone in single-threaded mode. The system was quantized with 120 bits per supervector, each supervector containing two log mel spectra, for an overall rate of 3 kb/s. The quantization was a two-dimensional vector-quantization of the KLT coefficients. Testing procedure To evaluate the absolute quality of the different systems on different SNRs a Mean Opinion Score (MOS) listening test was performed. Except for data collection, we followed the ITU-T P.800 [29] (ACR) recommendation. The data was collected using a crowd-sourcing platform with the requirements on listeners being native English speakers and using headphones. The evaluation dataset is composed of 30 samples from the Noisy VCTK dataset [30]: 15 clean and 15 augmented with additive noise at various SNRs (2.5, 7.5 and 12.5 dB). Each utterance for each system was rated about 200 times and the average and 95% confidence interval were calculated per SNR. Results The quality for the systems of Table 1 is shown in Figs. 1 and 2. The MOS with 95% confidence intervals are given for four SNRs. Fig. 1 displays the effect of predictive variance regularization and noise suppression (TN) without weight pruning and quantization. Predictive variance regularization results in a significant quality improvement and reduces the sensitivity to noise in the input signal. Noise suppression aids performance when noise is present. Fig. 2 shows the quality for pruned and quantized systems. For this case, the improvement due to variance regularization is particularly large for clean signals. The effect of noise suppression (TN) varies in an unexpected manner with SNR. This likely results from an interaction between noise suppression and quantization. It may be related to noise suppression reducing signal variability and quantization reducing noise on its own. As a reference, Fig. 2 provides the performance of the Opus codec [31] operating at 6 kb/s and the EVS codec [32] operating at 5.9 kb/s (for fairness with disabled DTX). It is seen that the proposed fully practical 3 kb/s WaveGRU coder performs significantly better than Opus at 6 kb/s and similarly to EVS operating at 5.9 kb/s. CONCLUSION We have developed a robust speech codec using neural-network based signal synthesis that encodes speech at 3 kb/s. Our system is suitable for, for example, low-rate video calls, and fits in consumer devices as evidenced by our implementation running on a wide range of mobile phones including the Pixel 2. Our experiments show that its quality is similar or better than state-of-the-art conventional codecs operating at double the rate. Our main contribution is that we addressed the impact of variability and distortion inherent in real-world input to practical speech codecs. We identified as causes for poor performance i) the inherent emphasis of outliers by the maximum likelihood criterion and ii) the difficulty of modeling a sum of multiple independent sources. We resolved these problems with predictive variance regularization and noise suppression.	2021-02-22T02:15:48.036Z	2021-02-18T00:00:00.000	{ "year": 2021, "sha1": "4b92f5b61dec03f7645d02387639f18cf6b8006e", "oa_license": null, "oa_url": "http://arxiv.org/pdf/2102.09660", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "4b92f5b61dec03f7645d02387639f18cf6b8006e", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Computer Science", "Engineering" ] }
261387941	pes2o/s2orc	v3-fos-license	Physicochemical and chemical properties of mung bean protein isolate affected by the isolation procedure The effects of different mung bean protein isolation methods on the chemical composition, the physicochemical properties, and selected antinutritional factors of mung bean protein isolates were investigated. Six protein isolates were prepared by isoelectric precipitation at different extraction pH levels (pH 8 and 9), by micellization, and by hybrid isolation at varying salt concentrations (0.25 M, 0.50 M, 0.75 M). The extraction conditions affected the amount of antinutritive compounds of the isolates. Compared to mung bean flour, micellization reduced phytic acid content by approximately 48% and trypsin inhibitor activity by around 88%. The remaining phytic acid concentration of the isolates influenced their re-solubility, particularly under acidic conditions. The protein isolates exhibited significant differences in surface hydrophobicity and thermal characteristics, indicating structural modifications caused by the extraction methods. Micellization and extraction at pH 8 were identified as mildest isolation methods, as evidenced by the highest enthalpy values. SDS-PAGE analysis demonstrated an enrichment of globulins and comparable protein profiles among the isolates, suggesting that the observed differences arise from conformational changes rather than variations in protein composition. The product yield in protein extraction from mung beans ranged from 8% to 19%, emphasizing the importance of enhancing overall extraction efficiency or exploring the utilization of by-products obtained during the protein isolation process. Introduction The world population is growing steadily.According to forecasts, more than nine billion people will be living on our planet in 2050 (United Nations, 2022).The associated overexploitation of natural resources is contributing to the global ecological crisis and climate change and impeding the nutrition of the entire world population.Excessive consumption of land and other resources for the production of animal proteins and an increasing consumer awareness with respect to health, nutrition, and the ecosystem has paved the way for growth in the alternative protein market.Pulses are an excellent source for the production of protein products due to their high protein content and their highly extractable proteins.The good extractability of legume proteins can be attributed to the protein storage anatomy in these crops, facilitating separation (Tenorio et al., 2018).Today, the most commonly used protein source among legumes is still soy.It is, therefore, of utmost importance to explore new alternative protein sources in order to provide a wider range of proteins with distinct functional properties for the replacement of animal protein.These include other legumes.The European Food Safety Authority (EFSA) has recently accepted mung bean protein isolate as a novel food (EFSA Panel on Nutrition et al., 2021).This opens up the possibility of using mung bean protein isolate as a plant-based protein alternative for numerous food products. Mung bean (Vigna radiata (L.) Wilczek) is used worldwide as a nutrient-rich food as well as an important animal feed.Mung beans are a rich source of proteins, fibers, and minerals (Mubarak, 2005;Shrestha et al., 2023).The protein content of mung bean ranges from 24% to 29% and is rich in essential amino acids, especially the aromatic amino acids leucine, isoleucine, and valine (Mubarak, 2005;Li et al., 2010).The amino acid content of mung bean protein isolate largely meets World Health Organization (1991) requirements, except for sulfur-containing amino acids and tryptophan (Kudre et al., 2013).Mung bean protein isolate was found to have good gelling and foaming properties and is well suited for texturization comparable to commercial soybean products (Li et al., 2010;Brishti et al., 2017).However, the presence of antinutrients can limit the applicability of mung bean and its isolate, since their occurrence can influence the bioavailability of proteins and minerals (Mubarak, 2005).Among others, phytic acid and trypsin inhibitors have been detected in mung bean (Mubarak, 2005).Several protein extraction techniques have been described that reduce the levels of antinutrients in various protein isolates from different sources such as soy or Moringa oleifera (Ali et al., 2010;Illingworth et al., 2022).Therefore, the choice of extraction technique can be an important factor to reduce the amount of antinutritional factors in the resulting protein isolate. Besides the chemical composition, isolation parameters strongly influence the technofunctional quality of a protein isolate.Wellestablished, conventionally used techniques for the production of protein isolates are alkaline extraction followed by isoelectric precipitation and salt extraction followed by dilutive precipitation, titled as micellization.Both procedures have already been performed on various beans and other legumes (Muranyi et al., 2013(Muranyi et al., , 2016;;Tanger et al., 2020;Illingworth et al., 2022).It was found that different pH values during alkaline extraction led to variations in protein yield, purity, and in the functional properties of the resulting protein ingredients (Das et al., 2021).In addition, high alkaline extraction values promote protein denaturation due to the strong pH-shift using isoelectric precipitation and, thus, affect protein solubility.Compared to isoelectrically prepared protein isolates, micellization resulted in protein isolates with less protein denaturation and higher protein purity, but lower protein yield, which was observed for various raw materials (Rahma et al., 2000;Muranyi et al., 2016;Illingworth et al., 2022).However, limited research has been conducted on the effects of micellization on the chemical and physicochemical properties of mung bean protein isolates (Shrestha et al., 2023), which is essential for the assessment of potential applications. It is already known that mung beans contain some antinutritional factors that can be reduced by dehulling.Therefore, dehulled mung beans were used in the present study.Until now alkaline extraction combined with isoelectric precipitation was studied on mung beans mainly at one extraction pH (Li et al., 2010;Brishti et al., 2017).Considering the relevance of milder isolation pH for the physicochemical and functional properties of protein isolates, a comparison of mung bean protein isolation at two different pH values was conducted in the present study.In addition, salt extraction was performed in combination with dilute precipitation to determine the effects of a milder isolation method on mung bean protein isolate.Salt extraction at 0.5 M NaCl has been studied previously (Rahma et al., 2000), but determination of the ideal NaCl concentration for mung bean protein extraction may enhance protein solubility, leading to improved extraction efficiency.Considering the distinct benefits of the two conventional isolation methods and their impact on technofunctional properties, the combination of both isolation techniques could provide hybrid isolates with high quality and yield.Therefore, salt extraction in combination with dilutive and isoelectric precipitation (= hybrid isolation) of mung bean proteins was performed in the present study for the first time.Different salt concentrations were utilized to produce hybrid isolates to examine the effect of NaCl during protein isolation.To evaluate the obtained protein isolates, the chemical composition and selected technofunctional properties and antinutritional factors were comprehensively investigated in our study.Thus, this study contributes to the advancement of isolation techniques for mung bean proteins by considering multiple factors such as pH, salt concentration, and hybrid isolation methods. Mung beans Commercial dehulled mung beans (Vigna radiata (L.) Wilczek) from Rapunzel Naturkost were purchased in Germany.The beans came from Chinese organic farming and were already mechanically shelled and halved.The beans were not heated above 40 • C at any time. Preparation of mung bean flour The seed kernels were ground with a blade-grinder (Grindomix GM200, Retsch, Haan, Germany) and sieved using a 300 μm mesh sieve. The flour was defatted with n-hexane at a 1:3 (w/v) ratio for 2 h and airdried for 24 h.The flour was stored at 10 • C until further use. Determination of isolation parameters by protein solubility curves To maximize protein yield for each isolation procedure, the protein solubility of mung bean flour was determined.The mung bean flour was diluted 1:10 (w/v) with water and the pH was adjusted from 2 to 10 with HCl (1 M, 5 M) or NaOH (1 M, 5 M).For salt extraction, the same procedure was done by dilution of the mung bean flour in salt solutions of 0.25 M, 0.50 M, and 0.75 M NaCl, respectively.The solutions were centrifuged at 1,000 g for 20 min (Heraeus™ Biofuge™ Stratos™, Thermo Fisher Scientific, Waltham, MA, USA).Protein content of the flour and the supernatant was analyzed via Dumas method (Dumatherm, Gerhardt, Germany).A nitrogen to protein conversion factor of 6.25 was used.The protein content of the samples was related to their respective dry weight (dw).The protein solubility was determined according to the Eq. ( 1): Preparation of alkali extracted and isoelectric precipitated mung bean protein isolates (IP) The protein isolation was performed according to the method of Brishti et al. (2017) with slight modifications.The defatted mung bean flour was diluted 1:10 (w/v) with water and the pH was adjusted to 9 (IP9) and 8 (IP8) using 1 M and 5 M NaOH.The suspension was stirred at room temperature for 1 h and centrifuged at 1,000 g for 30 min.The supernatant was acidified to pH 5 with 1 M and 5 M HCl and stirred for 30 min.The precipitate was recovered by centrifugation at 4,000 g for 25 min, was washed once with deionized water, and lyophilized using a freeze-drier (Model Alpha 2-4 LSC, Christ GmbH, Osterode, Germany). Preparation of micellized mung bean protein isolate (MP) The preparation of micellized protein isolate was done by the method of Muranyi et al. (2016) with slight modifications.The mung bean flour was diluted 1:10 (w/v) in an aqueous salt solution of 0.75 M NaCl with a native pH value of pH 6.1.After 1 h of stirring, the solution was centrifuged at 1,000 g for 30 min.The supernatant was diluted with deionized water at a ratio of 1:5 (to approx.0.125 M NaCl) and stirred for 30 min at room temperature.The dispersion was allowed to stand at 4 • C overnight and was centrifuged at 12,000 g for 25 min.The precipitate was washed with deionized water and lyophilized. Hybrid isolates (H): combination of salt induced extraction and dilutive isoelectric precipitation The defatted mung bean flour was diluted 1:10 (w/v) in aqueous salt solutions of 0.25 M (H0.25), 0.5 M (H0.50), and 0.75 M (H0.75)NaCl and the pH was adjusted to pH 6 with 1 M HCl according to the protein solubility curves.After stirring for 1 h, the dispersion was centrifuged at 1,000 g for 30 min.To induce dilutive precipitation, the supernatant was diluted again to approximately 0.125 M NaCl by dilution with deionized water at a ratio of 1:1, 1:3, and 1:5, respectively.The supernatant was precipitated isoelectrically by adjusting the pH to 4 with 1 M and 5 M HCl and stirred for 30 min.The dispersion was centrifuged at 12,000 g for 25 min, washed with deionized water, and freeze-dried. Protein yield and purity To assess the efficiency of the different isolation methods, the protein yield was determined according to Das et al. (2021) by the following Eq. Chemical composition The defatted mung bean flour and the protein isolates were analyzed for protein, starch, fat, moisture, and ash content.Protein content was determined by the Dumas method (Dumatherm, Gerhardt, Germany) based on the nitrogen content (N x 6.25).Starch content was measured enzymatically according to AOAC Method 996.11.The ash content was measured by combustion method at a temperature of 600 • C. To enhance the ashing process, a solution of magnesium acetate (12% w/v) was added and the mass of the magnesium acetate solution was subsequently measured and subtracted from the total ash of the samples.Fat content was determined according to AOAC Method 922.06 with slight modifications, instead of ethanol demineralized water was used.The moisture content was determined by heating the solid samples up to 100 • C (MA-30 Sartorius, Göttingen, Germany) until mass constancy and the dry weight was calculated gravimetrically.All values were expressed as a percentage of the respective dry weight of the samples. Determination of phytic acid The colorimetric determination of phytic acid was done enzymatically (Cat.No. K-PHYT, Megazyme International Bray, County Wicklow, Ireland) by measuring free phosphor and total phosphor content, respectively, resulting from acid extraction and treatment with Megazym phytase enzyme preparation.Depending on the predicted phytic acid content of the samples, 0.7-1.0g sample material was weighted and stirred overnight in 20 mL HCl (0.66 M).An aliquot of 1 mL of sample extract was centrifuged and 0.5 mL of the supernatant was mixed with 0.5 mL NaOH (0.75 M).For the determination of total phosphor content, 0.06 mL of distilled water, 0.2 mL of phytase assay buffer, and 0.02 mL of phytase enzyme preparation were added to 0.05 mL of sample extract and mixed accurately followed by incubation in a water bath at 40 • C for 10 min.Alkaline phosphatase buffer (0.2 mL) and alkaline phosphatase enzyme preparation (0.02 mL) were added, mixed well and placed in a water bath at 40 • C for 15 min.After 15 min, the reaction was stopped by adding 0.30 mL TCA (50% w/v) and a subsequent centrifugation followed (11,300 g, 10 min).The supernatant was used for colorimetric determination.The free phosphor content was determined by measuring the sample in absence of the addition of phytase and alkaline phosphatase enzyme preparation.Five phosphor standard solutions were prepared (0 μg, 0.5 μg, 2.5 μg, 5 μg, 7.5 μg) to calculate the mean value of phosphor standards.For the colorimetric determination, 1.0 mL of the samples supernatant and phosphor standards were mixed with 0.5 mL color reagent, incubated at 40 • C for 1 h and measured at 655 nm.Phytic acid content was calculated according to Eqs. (3)-( 5) and related to the respective dry weight of the samples: Determination of trypsin inhibitor activity The trypsin inhibitor activity (TIA) was measured according to the Trypsin Inhibitor Assay half volume method of Liu (2019).Enzyme extract was prepared by stirring 1 g sample material in 50 mL NaOH (10 mM) at room temperature for 3 h.Mung bean extract was diluted with deionized water to the extent that 1.0 mL of the extract caused trypsin inhibition of 30-70%.The test was performed in a water bath at 37 • C.An aliquot of 1.0 mL of the diluted extract was mixed with 2.5 mL Nα-benzoyl-DL-arginin-4-nitroanilid-hydrochlorid, 1.0 mL trypsin enzyme preparation, and 0.5 mL acetic acid (30%).Once the trypsin solution was added, the reaction was stopped after 10 min by the addition of 1.0 mL of 30% acetic acid solution.The mixture was centrifuged (14,100 g, 5 min) and measured at 410 nm (A410S).The reference value (A410R) was determined by measuring the reaction in absence of inhibitors by replacing the sample extract with water.Furthermore, reagent blank values for the sample measurements (A410SB) and reagent blank values for the reference values (A410RB) were carried out by addition of the acetic acid solution prior to the trypsin solution.For all absorbance measurements, deionized water was used as reference.A trypsin unit (TU) is defined as the increase in absorbance by 0.02 at 410 nm.Trypsin inhibitor units (TIU) are expressed per mg protein (dw) and calculated as follows: Color parameters and browning index For the color measurement, a Chroma Meter CR-400/410 (Konica Minolta, Osaka, Japan), equipped with a measuring head CR-400, a data processor DP-400, a CR-A44 white calibration plate, and glass cuvettes CR-A502 was used.The calibration was performed using a white standard with Y = 93.8,x = 0.3157, y = 0.3322 prior to color determination.The browning index was calculated using the equation of Brishti et al. (2020): Protein solubility of the protein isolates A 1% (w/v) suspension was produced by dispersing the respective isolate in distilled water.The solution was stirred for 1 h until a homogenous mixture was obtained.The pH was set from 2 to 10 with 1 M and 5 M NaOH or HCl.After centrifugation at 1,000 g for 30 min, the supernatant was analyzed for soluble protein content via Dumas method.A conversion factor of 6.25 was used. SDS-PAGE analysis SDS-PAGE analysis under reducing conditions (R) was conducted using a 12% precast polyacrylamide gel (Biorad Laboratories GmbH, Feldkirchen, Germany).A protein solution of 5 g/L was prepared by dispersing the samples in 500 μL of distilled water and by the addition of 500 μL Tris-HCl (0.13 M) buffer (pH 6.8) containing 1 g/L bromophenol blue, 20% glycerol (v/v), 60 g/L SDS, and 10% β-mercaptoethanol.The samples were mixed well, heated at 95 • C for 10 min, and were centrifuged at 14,000 g for 10 min.Aliquots of 5 μL of the supernatant were loaded into the gel chambers.The electrophoresis buffer (Biorad Laboratories GmbH, Feldkirchen, Germany) consisted of 25 mM Tris, mM glycine, and 0.1% SDS (pH 8.3) was diluted 1:10 with water.The electrophoresis was carried out at 120 V for approximately 1 h, until the tracking dye reached the bottom of the gel.The gel was stained with Coomassie Blue R-250 for 45 min and destained with a mixture of ethanol, acetic acid, and water (4:1:5, v/v).SDS-PAGE under nonreducing (NR) conditions were prepared by the same procedure, but without the addition of β-mercaptoethanol. Thermodynamic properties Thermodynamic properties were analyzed using a differential scanning calorimeter (DSC Q2000; TA Instruments, Lukens Drive, New Castle, DE, U.S.).The samples were prepared according to Devkota et al. (2023) with slight modifications.The isolates were suspended in demineralized water 20% (w/v), stirred for 2 h, and hydrated overnight.The measurement was done according to the method of Muranyi et al. (2016).The protein suspensions (10 mg) were heated in sealed standard aluminum pans.An empty pan was used as a reference.The pans were heated from 30 to 110 • C at a linear heating and cooling rate of 2 K/min.Enthalpy (ΔH) and denaturation temperature (T D ) were calculated by the TA Universal Analysis V.4.5A software (TA Instruments).The enthalpy values were related to the protein content (dw) of the samples. Surface hydrophobicity The surface hydrophobicity (S 0 ) was determined by the method of Nakai (2003) using 1-anilino-8-naphthalenesulfonate (ANS) with some adjustments.A 0.1% stock solution of protein isolate was prepared in phosphate buffer (10 mM, pH 7) and stirred at 4 • C overnight until the isolate had completely dissolved.The stock solution was diluted to five concentrations ranging from 0.004 to 0.06% (w/v).Then, 20 μL ANS solution was added (8.0 mM in 10 mM phosphate buffer, pH 7) to 1 mL of each solution.After stirring, the samples were left in the dark for min.Fluorescence measurement was done with a microplate reader (FLUOstar Omega, BMG Labtech, Ortenberg, Deutschland) at 355 nm λ excitation and 460 nm λ emission .The results were shown as the slope of the curve of protein concentration (dw) and fluorescence intensity.Measurement was performed with two independent samples measured in triplicate and corrected by a blank measured without ANS. Fourier transform infrared (FT-IR) spectroscopy The secondary structures of mung bean proteins were measured using a FT-IR spectrometer (Cary 630 FTIR, Agilent Technologies, California, USA).The spectra of the dry powder samples were recorded in the region of 4000-650 cm − 1 with a spectral resolution of 4 cm − 1 and scans.The FT-IR spectra were analyzed according to Fevzioglu et al. (2020) using OriginLab software Version 2023b (OriginLab Corporation, Massachusetts, USA).The empty crystal was used as background.For analysis, the original absorption spectra were normalized.The normalized spectra and the second derivative spectra were smoothed using a seven-point, third-degree polynomial Savitzky-Golay function.The normalized protein spectra within the amide I region (1600-1700 cm − 1 ) were used for the Gaussian curve-fitting process.Ideal alignment was achieved through iterative peak adjustment. Water (WAC) and oil absorption capacity (OAC) The water and oil absorption capacity were determined by the method of Brishti et al. (2017) with slight modifications.For the determination of WAC and OAC, 0.5 g sample material was dispersed in water at a ratio of 1:20 (w/v) and in corn oil in a ratio of 1:10 (w/v), respectively.The samples were left for 30 min, with remixing after 15 min to achieve a complete dispersion.After centrifugation at 5,000 g for 30 min, the water was decanted and the samples were weighted immediately.For OAC, the supernatant was decanted and the tube was left hanging at a 45-degree angle for 20 min to allow the oil to drain off.The WAC/OAC was calculated as follows: where W 1 is the weight of the tube plus dry sample [g], W 2 is the weight of the tube plus oil or water saturated sample, and W 0 is the weight of the dry sample [g].Protein content was related to the dry weight.The WAC or OAC is expressed as gram of water or oil absorbed per gram of protein isolate, respectively.To determine the samples relative water and oil absorbance, the water-oil absorption index (W/O index) was calculated based on the method of De Kanterewicz et al. (1987), calculated as follows: Statistics All results are shown as means of three replicates and the data were expressed as mean ± standard deviation.Statistical analysis of the results was performed using XLSTAT software (version 2022.3.1,Addinsoft Technologies, Paris, France).For pairwise comparisons, an ANOVA with Bonferroni post-hoc test (selected significance level p ≤ 0.05) was used. Determination of extraction parameters by protein solubility curves Analysis of the protein solubility curves.Protein solubility of mung bean flour at different pH values and different NaCl concentrations (Fig. 1) was evaluated to determine the isolation parameters.The extractability of proteins increased with higher pH and NaCl concentration.It is known that proteins have a negative net charge at alkaline pH, which increases their solubility due to stronger electrostatic repulsion.Therefore, the solubility without any NaCl addition was highest at pH (81.6%).Lowest solubility was found at pH 5 (10.9%).These findings were slightly higher compared to those provided by Wang et al. (2011) who found lowest solubility of mung bean protein at pH 4.4. The protein solubility further increased with rising salt concentrations (Fig. 1).This phenomenon is attributed to the salting-in effect, wherein proteins interact with salt ions and water molecules.The chloride and sodium ions interact with the oppositely charged protein groups causing the neutralization of their surface charges.Consequently, the electrostatic interactions between the proteins are reduced and their solubility is increased (Schröder, 2017).The addition of 0.75 M NaCl within the pH range of 6-7 resulted in the highest protein extractability, reaching levels of 83.5%-84.0%.As a result, protein solubility was significantly (p ≤ 0.05) increased by more than 15% (pH 7; 0.75 M NaCl) compared to alkaline extraction without NaCl (pH 7; M NaCl).In this way, high alkaline pH values can be avoided during extraction in order to achieve milder extraction conditions and high protein yields.Since the effects of salting-in and salting-out are depending upon the protein composition, charge differences, hydrophobic properties, and structure of the samples (Schröder, 2017), different protein materials may exhibit different behaviors. In addition, a shift of the isoelectric point from pH 5 to 4 was detected for salt supported extraction.Due to their smaller hydration radius negatively charged chloride ions can bind more strongly to positive charged protein groups than positively charged sodium ions to negatively charged protein groups.Therefore, the proteins become more negatively charged and repel each other more strongly at their original isoelectric point (Ockerman, 1996). Determination of extraction parameters.To analyze the influence of the extraction pH on the properties of the protein isolates, the extraction of mung bean protein was performed without any NaCl addition at pH and pH 9 for IP8 and IP9, respectively, and precipitation was carried out at the point of lowest solubility, pH 5 for both.Salt extraction of MP was performed with 0.75 M NaCl at a native pH of 6.1 and dilutive precipitation was done by adding water 1:5 (to approx.0.125 M NaCl).The resulting pH after dilution was 6.5.Hybrid isolates were extracted at different salt concentrations (0.25 M, 0.50 M, and 0.75 M) at pH 6.To induce dilutive precipitation, all salty protein extracts were diluted to a NaCl concentration of about 0.125 M by adding deionized water 1:1, 1:3, and 1:5, respectively.Considering the protein solubility curves at the different NaCl concentrations (0.25 M-0.75 M), the point of lowest solubility shifted from pH 5 to 4. Therefore, the solubility of mung bean proteins was determined at the salt concentration after dilutive precipitation (0.125 M) in order to maximize protein yield.The solubility at 0.125 M NaCl was found to be 19.0%and 20.4% at pH 4 and 5, respectively.Therefore, a precipitation pH of 4 was selected for the hybrid isolates. Chemical composition The chemical composition of mung bean flour and protein isolates is shown in Table 1.The protein content of the mung bean flour was 26.8% dw.Protein concentrations of mung bean protein isolates varied from 95.9% to 99.5% dw, for IP9 and MP, respectively, which were higher than reported in other studies of mung bean, where protein content ranged from 81.5% to 87.8% (Kudre et al., 2013;Brishti et al., 2017).Highest ash contents were found for MP (3.6% dw) and hybrid isolates (3.3%-3.4% dw) which is most likely due to the addition of salt during extraction (Table 1).Furthermore, the ash content increased with pH during alkaline extraction due to the addition of NaOH and HCl (Muranyi et al., 2013), as shown by the comparison of IP8 and IP9.The determined fat content of defatted mung bean flour was low (1.1% dw).A slight accumulation of fat was observed in IP8, IP9, and MP, while hybrid isolation technique marginally reduced the fat content. Protein and product yield Particularly with regard to economic aspects, product and protein yield are important factors for selecting the appropriate isolation method.Protein yields ranged from 30.4% to 67.5% dw for MP and IP9, respectively (Fig. 2).These results are comparable to those of Rahma et al. (2000), who found protein yields between 40.9% and 66.5% for micellized and isoelectric mung bean protein isolates, respectively.Product yield was lowest for MP (8.3%) and increased to a maximum of 18.8% for IP9.However, the product yield for all hybrid isolates was increased significantly (p ≤ 0.05) about 1.8 times compared to MP isolate. While protein yield increased with increasing pH, protein purity decreased slightly but significantly (p ≤ 0.05) with increasing pH in case of IP (Fig. 2).This decline in protein purity can be attributed to the higher solubility of non-protein components in alkaline pH regions (Das et al., 2021).Thus, there is a negative relationship between protein yield and protein purity, which has also been found in studies that have examined other protein sources like lupin, rapeseed, or amaranth (Tenorio et al., 2018;Das et al., 2021).Accordingly, micellar extraction provided the isolate with the highest protein purity but also with the lowest protein and product yield.In addition, an increase from 0.25 M to 0.75 M NaCl during extraction resulted in improved protein purity of hybrid isolates and slightly increased protein yields, consistent with Pickardt et al. (2009).Protein purity of H0.75 (0.75 M NaCl) is comparable to MP isolate, which makes this isolation procedure a potential method to obtain isolates with high purity. Antinutritional factors The mung bean flour exhibited a phytic acid content of 1.05 g/100 g dw sample (Table 1).This finding falls within the range of phytic acid contents observed in various mung bean genotypes and varieties, which typically range from 0.57 to 1.90 g/100 g (Dhole and Reddy, 2015).The isolation methods employed in this study had a strong impact on the phytic acid content.In particular, isolates IP8, IP9, and H0.25 to H0.75 showed an accumulation of phytic acid, while the MP isolate exhibited a reduction.Among the isolates, the highest phytic acid content was observed for H0.25 to H0.75, which had approximately three times higher phytic acid content compared to the defatted mung bean flour (Table 1). Phytic acid is soluble in a wide range of pH and mostly negatively charged at pH values found in foods because of the six phosphate groups with different pK a values (Wang and Guo, 2021).With increasing pH, the negative charge of phytic acid increases, so that the interaction with negatively charged proteins is reduced in alkaline pH ranges.However, indirect linkage of protein and phytate can still occur through complex formation via multivalent cations (Wang and Guo, 2021).When the pH is decreased to the isoelectric point, the proteins negative net charge is neutralized, leading to a reduction in repulsive forces.As a result, a greater number of electrostatic interactions can occur, leading to the formation of more protein-phytate complexes.By comparing MP and hybrid isolates it can be seen, that mainly the precipitation method had an influence on the phytate content (Table 1).As MP was consistently extracted and isolated in the region of pH 6, where the net charge of the proteins is more negative, less phytic acid was bound.Precipitation of hybrid isolates at pH 4 favored the interaction between the proteins and phytic acid.Considering IP8 and IP9, alkali extraction and isoelectric precipitation at pH 5 resulted in lower phytic acid levels than hybrid isolation.Values are means ± standard deviation (n = 3).Abbreviations: IP = alkali extracted and isoelectric precipitated mung bean protein isolate at pH 8 (IP8) and pH 9 (IP9); MP = micellized mung bean protein isolate; H = hybrid isolate prepared with 0.25 M (H0.25), 0.50 M (H0.50), 0.75 M (H0.75)NaCl. Furthermore, it seems that an increase in salt concentration from 0.25 M to 0.75 M NaCl had a slightly reducing effect on the phytate content in the different isolates.An increase in salt concentration generally hinders the protein-phytate complexation (Wang and Guo, 2021).Therefore, a small quantity of phytic acid could be separated after the first centrifugation step during hybrid isolation.However, the phytate reducing effect of NaCl at 0.75 M was marginal (Table 1) compared to H0.25 and H0.50 for effective phytic acid reduction. The mung bean flour exhibited a trypsin inhibitor activity (TIA) of 20.51 TIU/mg protein dw (Table 1).This value was higher compared to literature data obtained by Mubarak (2005) where TIU/mg protein of dehulled mung bean was 14.6 TIU/mg protein.The isolation method had a strong influence on the TIA of the isolates.Compared to the flour, alkali-extracted protein isolates showed a reduction in TIA by almost half.The increase from pH 8 to pH 9 in the extraction medium had a slightly decreasing effect on TIA.Micellization and hybrid isolation were able to reduce TIA by up to 90% compared to the flour.It can be assumed that the extractability of trypsin inhibitor at pH 6 is reduced and, therefore, trypsin inhibitor was successfully separated after the first centrifugation step for micellar and hybrid isolates.In a study of Chrispeels and Baumgartner (1978), the isoelectric point of trypsin inhibitor from mung bean was defined at pH 5.05, which supports the previous thesis. Color analysis With the exception of MP and the flour, all samples showed positive values for a* and b* (Table 2).The high b* values indicate a yellow hue with a slight reddish hue due to small a* values, which is consistent with the findings of Brishti et al. (2020).MP exhibited a slightly greenish hue expressed by the negative a* value, most similar to the flour.The highest L* values were observed for MP and the hybrid samples.Under alkaline conditions, the extraction of polyphenols is promoted, leading to the formation of brown quinones that can react with proteins, resulting in brown coloration (Xu and Diosady, 2002).This is possibly the reason for the higher browning index of the IP isolates compared to MP and the hybrid isolates (Table 2).Therefore, the lower extraction pH of micellar and hybrid isolation method had a positive impact on the brightness of the mung bean protein isolates.Similar effects were observed in sunflower protein isolate, where lower pH values in NaCl-assisted extraction prevented the covalent binding of phenolic compounds to proteins (Pickardt et al., 2009). SDS-PAGE analysis Mung bean storage proteins mainly consist of 60% globulins and 25% albumins (Yi-Shen et al., 2018).The predominant storage protein found in mung beans is the 8S globulin, also known as vicilin.It is estimated to make up approximately 89% of the total globulin content in mung beans.The remaining globulin fraction is composed of the 11S globulin (8%) and the 7S globulin (3%) (Mendoza et al., 2001).SDS-PAGE (Fig. 3) was performed to analyze the changes in protein composition among the different protein isolates.Conventional alkali extraction and isoelectric precipitation is known to accumulate globulins in the resulting protein isolate, since the isoelectric point of globulins is mostly between pH 4 to 5, whereas albumins have a broad solubility profile over a pH range from 2 to 12 (Tanger et al., 2020, Yang et al., 2023).The protein isolates examined in this study demonstrated an accumulation independent of the isolation technique applied and showed comparable band patterns.Six bands were predominantly present under non-reducing conditions: two bands with a molecular weight of 24 and 28 kDa, respectively, a high intensity band ranging from 43 to 50 kDa, and three bands at 57, 62 and 70 kDa, respectively.Four bands (24,28,(43)(44)(45)(46)(47)(48)(49)(50)and 62 kDa) showed no alterations when compared to the SDS-PAGE under reducing conditions, indicating that these bands likely correspond to the 8S globulins, which lack in disulfide bridges (Mendoza et al., 2001).Under reducing conditions, the band at 70 kDa Abbreviations: IP = alkali extracted and isoelectric precipitated mung bean protein isolate at pH 8 (IP8) and pH 9 (IP9); MP = micellized mung bean protein isolate; H = hybrid isolate prepared with 0.25 M (H0.25), 0.50 M (H0.50), 0.75 M (H0.75)NaCl. a Flour was not included in the Anova to focus on the variations between the different isolates.disappeared, while a band at 20 kDa emerged.This band at 20 kDa can be assigned to 11S globulin, which is consistent with the findings of Liu et al. (2015).Mendoza et al. (2001) identified another subunit of 11S globulin at 40 kDa, which could be included in the high-intensity band around 43 kDa in the present study (Fig. 3).Micellization can result in protein isolates consisting of a mixture of globulins and albumins, since the proteins are isolated by micelle formation rather by their isoelectric point (Tanger et al., 2020).Under reducing conditions, characteristic albumin bands from mung bean were identified by Yang et al. (2023) at 25 kDa and between 70 and 95 kDa.In the present samples, bands at 25, 80, and 95 kDa were detected indicating the presence of albumin, although they were less dominant compared to the globulin bands.This demonstrates that some albumins were present in the different isolates.However, despite isolation by micellization, an albumin accumulation could not be observed in MP.The protein patterns in all isolates were comparable, suggesting that the techniques did not exhibit selectivity for specific globulin or albumin fractions. Protein solubility of the protein isolates Protein solubility of the final protein ingredients influences their emulsifying activity, foaming capacity, and water holding properties (Deng et al., 2011).Therefore, solubility characteristics are highly important for technological application of protein isolates in food products.As illustrated in Fig. 4, the isolation procedure had an impact on the protein solubility.In the alkaline range above pH of 8, all isolates exhibited a solubility higher than 87%.The lowest solubility of all isolates was observed in the range from pH 4 to pH 6, which is consistent with previous studies on mung beans (Rahma et al., 2000;Brishti et al., 2017).The comparison between IP8 and IP9 revealed that IP8 had a significantly (p ≤ 0.05) higher solubility, particularly between pH 3 and 6, than IP9, which can be attributed to the lower degree of denaturation, due to less harsh extraction conditions (Tanger et al., 2020).IP8 showed the lowest protein solubility at pH 5 (17.0%) and IP9 between pH 4 and 6 (12.3-19.0%)(Fig. 4). Comparing the solubility profiles of MP, IP, and hybrid isolates, MP revealed the highest protein solubility profile, with an isoelectric point at pH 6, where solubility was the lowest at 14.2% (Fig. 4).There was a shift in the isoelectric point from pH 4 to pH 6, for H0.25, H0.50, H0.75 to MP respectively.This shift can be attributed to the decreasing phytic acid concentration in these samples, as discussed in Section 3.4.Ali et al. (2010) detected a relationship between reduced phytic acid content and enhanced solubility of soybean protein within the pH range from 2 to 4.5.This improved solubility can be attributed to a reduction in the formation of insoluble protein-phytate complexes.The width of the solubility curves for hybrid isolates was comparable to that of the IP isolates.Among the different hybrid isolates, solubility improved with increasing salt concentration during extraction.In particular, at acidic pH values of 3 and 6, H0.75 showed significantly (p ≤ 0.05) higher solubility compared to H0.25 (Fig. 4).However, at pH 2 and 7, the solubility of H0.75 was equivalent to that of H0.25. Thermal properties The effects of the different isolation techniques on the thermal stability and denaturation state of the protein isolates were determined via DSC measurement.Thermal properties like denaturation temperature (T D ) and enthalpy of denaturation (ΔH) are presented in Table 3.The denaturation temperature (T D ) represents the thermal stability of the protein isolate, while ΔH represents the energy required to initiate the denaturation process.These parameters are influenced by factors such as protein structure and conformation, protein content, pH, and ionic strength (Arntfield and Murray, 1981;Murray et al., 1985). The T D values ranged from 82.9 to 86.1 • C for IPs and MP, respectively, whereas enthalpy of denaturation was significantly (p ≤ 0.05) highest for MP and IP8 (Table 3).Higher enthalpy values indicate a more native state of protein (Tanger et al., 2020).Micellization is known as a milder isolation method than alkaline extraction and isoelectric precipitation, leading to less denaturation and more ordered globular structures, which can be associated with the higher ΔH values (Muranyi et al., 2013).Protein isolates from cowpea and pigeonpea demonstrated that micelle isolates had higher enthalpy values than isoelectric prepared protein isolates, but not necessarily higher denaturation temperatures (Mwasaru et al., 1999).High T D values are associated with high thermal stability and more hydrophobic interactions (Arntfield and Murray, 1981).The enhanced thermal stability observed in IP8 and IP9 could potentially be attributed to partial unfolding that occurred during the isolation process.Furthermore, the enthalpy of denaturation for IP8 and IP9 increased as the extraction pH decreased, aligning with findings from previous studies on amaranth, cowpea, and pigeonpea (Mwasaru et al., 1999;Das et al., 2021).In general, an increase in extraction pH lowers ΔH as described by Arntfield and Murray (1981) due to increased unfolding and denaturation of proteins.The transition enthalpy of the hybrid isolates was slightly higher than for IP9, but significantly (p ≤ 0.05) lower than for IP8 and MP ranging from 4.6 to 7.2 J/g protein dw (Table 3).However, no significant (p ≤ 0.05) differences were observed among the hybrid samples regarding T D and ΔH.Consequently, it can be concluded that increasing the salt concentration from 0.25 to 0.75 M did not provide additional stabilization of the protein structure and that precipitation at pH 4 led to structural and conformational changes of the hybrid isolates. Surface hydrophobicity Surface hydrophobicity (S 0 ) represents the number of hydrophobic groups located on the surface of the proteins.On the one hand, denaturation of proteins leads to partial unfolding and exposes hydrophobic groups on the protein surface, which increases surface hydrophobicity (Kato and Nakai, 1980).On the other hand, increasing denaturation leads to the association of these hydrophobic groups, which decreases surface hydrophobicity (Wang et al., 2014).Therefore, S 0 helps to visualize the state of aggregation and denaturation of protein samples.In addition, surface hydrophobicity can be related to technofunctional properties such as protein solubility, emulsifying capacity, as well as oil absorption capacity (Deng et al., 2011). The surface hydrophobicity of mung bean protein isolates at pH 7 ranged from 3267.7 to 4613.2 for H0.25 and IP9, respectively (Table 3).When comparing the alkaline extracted and isoelectric precipitated protein isolates (IP 8 and IP9) with the micellar protein (MP), it becomes evident that MP had a significantly (p ≤ 0.05) lower surface hydrophobicity compared to IP8 and IP9.This suggests that the hydrophobic groups of MP are primarily located inside the protein structure.Muranyi et al. (2013) reported the formation of micellized lupin protein with a hydrophilic shell and a hydrophobic core during dilutive precipitation, which is achieved by arranging hydrophobic regions inward to attain the most energetically favorable state.In contrast, the extraction conditions were harsher for IP8 and IP9.The abrupt transition from the extraction pH to the precipitation pH of 5 did not provide enough time for the proteins to properly position their hydrophobic groups towards the interior.For this reason, IP9 may have more hydrophobic groups exposed on its surface than IP8 and MP.The surface hydrophobicity values of the hybrid isolates were affected significantly (p ≤ 0.05) by the different salt concentrations (Table 3), but did not appear to follow a consistent trend.Muranyi et al. (2013) reported that the protein structures resulting from the combination of salt extraction and isoelectric precipitation of lupin protein isolates consist of a mixture of micellized proteins and isoelectric irregular protein aggregates.Therefore, the surface hydrophobicity of H0.25, H0.50, and H.075 could be attributed to the presence of both micellized structures and isoelectrically agglomerated structures.The isolation method and varying NaCl concentrations had a significant (p ≤ 0.05) impact on the surface hydrophobicity of the isolates, thereby impacting the conformation of the proteins. Fourier transform infrared spectroscopy The composition of the secondary structure is presented in Table 3. Supplementary information (Fig. S1) provides comprehensive details on the second deviation and the Gaussian curve-fitting.The FT-IR spectra of the protein isolates were similar with slight variations in the absorption intensities.Peaks were observed in the following ranges: 1628-1632 cm − 1 , linked to β-sheet structures; 1658 cm − 1 , assigned to α-helix and random structures; and 1684-1688 cm − 1 , defined as β-turn structures (Fevzioglu et al., 2020).The detected peak in the frequency range of 1660-1640 cm − 1 was assigned to α-helix and random structures, due to the overlapping peaks of α-helix and random structures in this region (Fevzioglu et al., 2020).All samples had high contents of β-sheet, α-helix, and random structures.MP isolate exhibited the significantly (p ≤ 0.05) highest percentage of β-sheet (48.4%) and lowest percentage of α-helix and agglomerated structures (46.1%) compared to all other isolates, which is mostly attributed to the milder extraction process and is consistent with the findings for surface hydrophobicity and denaturation enthalpy in the present study.According to Mune Mune et al. ( 2017), the β-turn structures are product of protein unfolding of higher ordered structures.The highest percentage of β-turn structures was found for IP9 (6,9%) which is also in line with the findings for denaturation enthalpy.However, percentages of secondary structure of IP8, IP9, and hybrid isolates showed no significant (p ≤ 0.05) differences, indicating that the different pH values and salt concentrations during extraction did not have a significant effect on the secondary structure. Water and oil absorption capacity The water and oil absorption capacity of a protein isolate determines its technofunctional properties and plays an important role in food formulations.Both water and oil absorption capacity influence food texture due to fat and water retention and are correlated to emulsion capacity and stability (De Kanterewicz et al., 1987;Elizalde et al., 1988).High WAC is important to bind water in food formulations, which in turn influences thickening and viscosity in dough, whereas OAC is important for fat entrapment in meat and flavor retention (Kinsella, 1979). WAC ranged from 1.6 to 2.3 g water/g protein dw for H0.25 and IP9, respectively (Table 3).The highest WAC was observed for IP9 and IP8, indicating a greater availability of hydrophilic amino acids for interaction with water.A higher WAC has also been associated with a higher degree of protein denaturation due to the increased surface to mass ratio of denaturized protein and, thus, with a lower protein solubility (Muranyi et al., 2016).As discussed in Section 3.8, isolation at pH of 8 instead of 9 resulted in a less denaturized sample which had a slight effect on the WAC of the both IP isolates.The same effect was seen for MP, which had lower WAC compared to IPs (Table 3).However, WAC of the hybrid isolates did not differ significantly (p ≤ 0.05) from MP isolate, indicating that the WAC is influenced by multiple factors. The OAC of the isolates ranged from 1.4 to 1.9 g oil/g protein dw for H0.25 and IP9/IP8 (Table 3), respectively.OAC was the highest and quite similar for IPs and MP, while a significant (p ≤ 0.05) reduction was observed in hybrid isolates.A high content of β-sheet is related to higher OAC values, which is in line for MP (Mune Mune et al., 2017).However, also a low degree of denaturation enhances protein flexibility and improves the ability of proteins to align their hydrophobic domains to the oil phase (De Kanterewicz et al., 1987;Muranyi et al., 2016).This could be the reason for higher OAC values of IP8.However, IP9 was the most denaturized sample with comparable OAC values to IP8 and MP.As discussed in Section 3.9 and 3.10, sample IP9 had the highest values for surface hydrophobicity and the highest percentage of β-turn structures.Therefore, the OAC value of IP9 could be explained by the partial unfolding and exposure of previously hidden hydrophobic regions to the proteins surface.This in turn promoted the interaction between lipids and proteins and resulted in a higher OAC value (Mune Mune et al., 2017).Increasing NaCl concentration during salt-induced extraction from 0.25 M to 0.75 M resulted in a slight significant (p ≤ 0.05) increase in OAC. A balanced ratio of hydrophilic and lipophilic properties of protein isolates contribute to improved emulsifying properties.Therefore, De Kanterewicz et al. (1987) calculated the W/O index and found a correlation to the emulsion capacity.The emulsion capacity is higher in the range of two, when a protein can absorb twice as much water as oil.It has to be considered that the W/O index is not directly attributed to the emulsion stability, which is more influenced by the amount of hydrophobic and hydrophilic groups of the protein.The W/O index for analyzed samples in our study ranged from 0.9 for MP to 1.2 for IP9 and H0.25 isolates (Table 3), indicating that the balance between hydrophilic and lipophilic properties is more on the hydrophilic side, except for the MP isolate.The W/O indices of our samples are comparable to bean protein isolate analyzed by De Kanterewicz et al. (1987) with a W/O index of 1.3.Therefore, IP9 and H0.25 would probably be the best choice to achieve a good emulsion capacity.The different salt concentrations during preparation of hybrid isolates did not significantly (p ≤ 0.05) change the W/O index. Conclusions Alkali extraction of mung bean proteins at pH 9 and their isoelectric precipitation at pH 5 resulted in protein products with the highest protein yield of 67.5%.The addition of NaCl during extraction significantly increased the extractability of mung bean protein in the saltassisted isolation procedures.However, the resulting salt concentration after dilution decisively determines the yield of precipitated protein, since part of the proteins is kept in solution by the salt.Therefore, dilution in combination with isoelectric precipitation did not increase the protein yield of the hybrid isolates beyond 54.8%.The product yields obtained from the different isolation methods ranged from 8 to 19%, highlighting the need for identifying potential utilization of the side products.This is crucial for developing sustainable and economically viable protein isolation methods.All isolation methods yielded proteinrich isolates with protein contents exceeding 95%.Micellization had the highest potential to generate a highly functional protein isolate, as it effectively reduced the amount of antinutritive compounds and improved the re-solubility of the protein isolate.However, the low product yield of micellized protein isolate is an economic drawback for large-scale applications.Lowering the extraction pH from 9 to 8 resulted in a less denaturized protein structure, as determined by DSC measurements, and improved solubility behavior of the alkali extracted proteins.As hypothesized, salt extraction in combination with dilutive and isoelectric precipitation, combined some attributes of both conventional isolation techniques.Therefore, hybrid isolation increased protein yield compared to micellization while producing an isolate with lighter color and lower trypsin inhibitor activity than the alkali extracted protein isolate.The surface hydrophobicity of the hybrid isolates suggested the occurrence of structural modifications due to extraction performed at varying NaCl concentrations.However, only micellization showed significant differences in the secondary structure.Regarding antinutritional factors, alkaline extraction with isoelectric precipitation and hybrid isolation were not sufficient to reduce the phytic acid content.Additional steps, such as acid pre-extraction or germination, should be considered to produce mung bean protein isolates with low phytic acid content.This study highlights the importance of considering structural changes, antinutritive factors, solubility and preservation of the proteins native state when selecting an appropriate extraction method.Further investigations of emulsifying, foaming, and gelling properties are necessary to comprehensively evaluate the technological applications of the protein isolates. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Fig. 1 . Fig. 1.Protein solubility profile of mung bean flour at different pH-values and different NaCl concentrations.Values are means ± standard deviation (n = 3). C .Wintersohle et al. Abbreviations: IP = alkali extracted and isoelectric precipitated mung bean protein isolate at pH 8 (IP8) and pH 9 (IP9); MP = micellized mung bean protein isolate; H = hybrid isolate prepared with 0.25 M (H0.25), 0.50 M (H0.50), 0.75 M (H0.75)NaCl; TIA = Trypsin inhibitor activity; TIU = Trypsin inhibitor units.aAll values are related to the dry weight of the respective sample.b Flour was not included in the Anova to focus on the variations between the different protein isolates. Table 1 Major components and antinutritional factors of mung bean flour and protein isolates.Different letters indicate significant differences (p ≤ 0.05).Values are means ± standard deviation (n = 3). Table 2 Color parameters of mung bean flour and protein isolates.Different letters indicate significant differences (p ≤ 0.05).Values are means ± standard deviation (n = 3). Table 3 Physicochemical and structural properties of mung bean protein isolates obtained by different isolation methods.Different letters indicate significant differences (p ≤ 0.05).Values are means ± standard deviation (n = 3). aValues are related to the dry weight of the respective sample.C.Wintersohle et al.	2023-08-31T15:03:22.824Z	2023-08-01T00:00:00.000	{ "year": 2023, "sha1": "9a96dfe56175b4bdaf68a97a5c37b73f328bf82d", "oa_license": "CCBYNCND", "oa_url": "https://doi.org/10.1016/j.crfs.2023.100582", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "f7a4e61cb828bba1f116afbdd52e74d8b6bae3c0", "s2fieldsofstudy": [ "Agricultural And Food Sciences" ], "extfieldsofstudy": [ "Medicine" ] }
252977348	pes2o/s2orc	v3-fos-license	Subtotal Pleurectomy with Intrathoracic Chemo Hyperthermia (HITHOC) for IVa Thymomas: De Novo Versus Recurrent Pleural Disease Simple Summary Stage IVa thymomas are rare entities in thoracic oncology without a standard of care. Evidence-based guidelines for the management of located pleural carcinomatosis are lacking. Surgery when feasible has an excellent prognostic factor. However, the technical choice is vast, from extra pleural pneumonectomy with high rates of morbidity and/or mortality to debulking with high rates of relapse. We investigated parietal subtotal pleurectomy combined with HITHOC in highly selected patients. The goal was to determine overall survival (OS) and the disease-free interval (DFI). Our findings suggest a real impact of this surgical procedure in distant relapse (DR) or de novo tumors (DNT). In this orphan disease, prospective and randomized trials are needed to clarify the place of HITHOC in the multimodal modern care of these patients. Abstract Introduction: Stage IVa thymoma is a rare disease without a standard of care. Subtotal pleurectomy and HITHOC introduced in highly selected patients may provide interesting oncologic results. The purpose of this study was to distinguish de novo stage IVa tumors (DNT) from distant relapse (DR) with respect to post-operative and long-term outcomes to provide the procedure efficacy. Methods: From July 1997–December 2021, 40 patients with IVa pleural involvement were retrospectively analyzed. The surgical procedure was subtotal pleurectomy and HITHOC (cisplatin 50 mg/m2, mitomycin 25 mg/m2, 42 °C, 90 min). The post-operative outcome, disease-free interval (DFI) and overall survival (OS) were analyzed. Results: Mean age was 52 ± 12 years. B2 and B3 thymomas were preponderant (27; 67.5%). The median number of pleural nodes were nine (4–81) vs. five (1–36); p = 0.004 * in DNT and DR, respectively. Hospital mortality rate was 2.5%. There were four specific HITHOC complications (10%). DFI were 49 and 85 months (p = 0.02 ), OS were 94 and 118 months (NS), in DNT and DR, respectively. Conclusions: Subtotal pleurectomy with HITHOC in IVa offers satisfying results in highly selected patients, for both DNT and DR. Due to the disease rarity, multicentric studies are needed to define HITHOC as a standard of care. Introduction Thymomas arising from thymic epithelial cells are the most frequent anterior mediastinal tumors belonging to orphan thoracic oncology [1] with a French incidence of about 250-350 cases per year [2]. The association of tumoral thymic epithelial cells and normal lymphocytes respecting the normal architecture of the thymus characterizes thymomas. In most cases, thymomas present indolent behavior with a tendency of loco-regional invasion. However, in 5% of cases, thymomas can spread along serous membranes as pleura (70%) and pericardium, rather than lymph nodes or distant organs [3]. When feasible, surgical en bloc resection R 0 is one of the most significant prognosis factors in the multimodal assessment therapy of thymomas [4,5]. In more detail, if surgery is performed in the early stages, surgery can be associated with postoperative radiotherapy following ITMIG recommendations, whereas neo-adjuvant chemotherapy is used for voluminous tumors with reasonable doubts of feasibility of initial R 0 surgery [6]. Finally, the association of chemotherapy and radiotherapy is the standard association in unresectable tumors. Masaoka's stage IVa is characterized as a tumor with microscopically confirmed separate nodes from the primary tumors involving the pleura (parietal and/or visceral) or pericardial and/or epicardium surface [7]. Stage IVa is generally identified during oncological workup by means of an Iodine-CT-Scan and/or 18-FDG-Pet-Scan, but can be accidentally discovered during thymic surgery. The curation of IVa includes multimodal strategies and innovations. Multimodal strategies have been purposed as combined chemo-radiations or surgery. However, the exact role of surgery is still unclear and controversial. Surgical options range from debulking (removal of 90% of entire tumors) to localized pleural surgery until extra-pleural pneumonectomy (EPP). The particularity of stage IVa is that it is a metastatic disease where R 0 surgery is not feasible. Secondary to a controversial debulking surgery or the high rate of mortality/complications of EPP, some authors in the 1990's advocated for a new concept derived from Sugarbaker experience for peritoneal/pleural malignant mesothelioma [8], introducing pleurectomy/decortication combined with intrathoracic chemo hyperthermia to cure stage IVa thymomas. The concept is based on a two-step surgery: (1) Subtotal pleurectomy associated with wedges and phrenic muscle resection where all visible tumor is removed, but in essence, the same as R 1 and (2) A hyperthermic intracavitary cisplatinbased chemotherapy to destroy the microscopic tumor residue [9,10]. Due to the rarity of the disease, only a small number of specialized surgical teams can present retrospective studies evaluating the place of HITHOC in stage IVa thymomas. In highly selected patients, HITHOC may provide better results in progression-free survival and overall survival; however, in a rare oncologic disease, the difficulty lies in the identification of the sub-group to enhance results. In this setting, based on our experience in HITHOC, we retrospectively analyzed our results and introduced a comparison between the following two sub-groups: IVa de novo (DNT) and IVa distant relapse (DR) pleural spread. Study Design From July 1997 to December 2021, we retrospectively analyzed the prospectively collected data of patients with pleural involvement of thymomas who underwent cytoreductive subtotal pleurectomy followed by intrathoracic chemo hyperthermia (HITHOC). The study was conducted in the Department of Thoracic Surgery, Lung and Heart-Lung Transplantation, Louis Pradel Hospital, a university hospital of the Hospices civils de Lyon, and a French expert center for thymic malignancies labialized by the national RYTHMIC network (thymic tumors and cancer). Data were acquired from a prospective database and follow-up was collected from secured medical charts (Easily ® medical software, Hospices Civils de Lyon, France). According to French laws, the study protocol was approved and registered by the ethical committee of the SFCTCV (French society of thoracic and cardiovascular surgery, IRB: 00012919; number 2022-04-06-19470). Aims of the Study The primary end-point of the current study was to evaluate, in a large series of a rare tumoral disease, the results of HITHOC. The secondary end-point was to determine the role of oncological outcomes on de novo IVa (DNT) vs. distant relapse (DR), with special attention focused on disease-free interval (DFI) and overall survival (OS). Preoperative Evaluation All the patients underwent standard preoperative assessment for major thoracic surgery with complete preoperative history and thymic oncological history and physical examination. Morphological and oncological evaluation were realized with standard chest X-ray, iodine CT-Scan, 18-FDG-Pet-Scan. MRI was performed when there was suspected major vessel tumoral invasion. A specific cardiac evaluation was performed with echocardiography and stress test when there were major cardiovascular risk factors such as previous chemotherapy with CAP and mediastinal radiotherapy. Respiratory function tests were assessed. Finally, a complete blood count, liver function test, and renal function tests (including creatinine clearance tests) were conducted. From 1997 to 2012, according to French guidelines, all patients were presented to an institutional medical board including surgeons, onco-pneumologists, and radiotherapists. From 2012 to the present, the national board of the RYTHMIC network reviewed all patients. Inclusion criteria are defined elsewhere [11]. Surgical Technique and Devices The protocol for pre-hydration, initially assessed by a mean of 2-3 L of saline serum, was stopped in 2018 and replaced by a specific protocol of nephro-protection by means of natrium thiosulfate in analogy of cisplatin nephro-protection used in intra-peritoneal chemo hyperthermia [12,13]. Natrium thiosulfate was intravenously released just before endocavitary chemotherapy at 9 mg/m 2 over 20 min, then a new injection at 12 mg/m 2 was instilled for 6 h immediately after surgery. All patients with myasthenia gravis (MG) were prepared for surgery with the intravenous perfusion of human immunoglobulin (Tegelin ® , LFB Biomedicaments, France) to prevent a post-surgical acute MG crisis. The HITHOC procedure was conducted under general anesthesia after placement of a thoracic epidural catheter. Venous and arterial monitoring was continuously assessed using internal jugular central and radial artery catheterization. Internal temperature was controlled using a bladder thermic catheter. All patients were positioned in the lateral decubitus position. A 5th or 6th posterolateral thoracotomy without serratus preservation was assessed (to avoid detachment spaces), and an initial extra-pleural dissection was performed. A subtotal parietal and phrenic pleurectomy was performed and lung or associated phrenic muscle resection were conducted when necessary [14] to achieved complete macroscopic tumor removal. After subtotal pleurectomy, parietal argon beam electrocoagulation was performed. Two 24 F-chest tubes were inserted and the thoracotomy was hermetically closed. The tubes were finally connected to the HITHOC device. The patient was re-positioned into a dorsal decubitus position and endocavitary chemotherapy was performed. The HITHOC device was initially made of a local designed material (Cavitherm, Soframedical, Vienne, France), replaced in 2018 by the Sunship2 (Gamida, Eaubonne, France). Antimitotic agents were cisplatin 50 mg/m 2 and mitomycin 25 mg/m 2 perfused endocavitary at 42 • C with 1.4 L/m 2 of isotonic saline solution over 90 min at 600-800 mL/min, according to central venous tension (<12 mmHg). This chemotherapy regimen was initially established following previous experiences in malignant pleural mesothelioma. At the end of the procedure, patients were transferred to the intensive care unit for tube removal and cardiothoracic monitoring. Follow-Up A complete follow-up was conducted using clinical and morphological data. Patients were routinely seen at 1, 3, and 6 months after discharge. Routine yearly CT scans were performed. Adverse events (AEs) were recorded using the National Cancer Institute Grading. Hospital mortality and morbidity were recorded within a <30 days post-surgery interval. Follow-up was obtained for all cases and censored on 31 December 2021. Statistical Analysis Statistical analysis was performed using GraphPad Prism (GraphPad) software. All tests were done using a significant threshold of α = 0.05. Overall survival (OS) and diseasefree interval (DFI) were defined from the date of HITHOC and last follow-up of death or evidence of disease recurrence within the study period. OS and DFI were estimated using Kaplan-Meier's method. Group comparisons were made using with the t-test or Mann and Whitney test. Multivariate analysis was unable to be performed due to the small sample size. Intra-Operative Results Of the total population studied (n = 40), 24 patients (60%) underwent associated lung wedge resections and 24 (60%) underwent partial phrenic muscle resections with direct nonresorbable stitches reinforced with Teflon ® patches. The median number of resected nodes was five, with a range of 1-81. The median number of resected nodes with significantly higher in the DNT group (9 vs. 5 ; p = 0.004). Per-operative surgical resection results are shown in Table 2. Four patients (10%) underwent extra-pleural pneumonectomy with HITHOC procedure due to previous surgical pleural symphyses. Seventeen (42.5%) patients had redux pleural surgery due to associated lobectomy (n = 3; 7.5%), superior vena cava replacement (n = 3; 7.5%), and wedge resections to upper lobes (n = 11; 27.5%) during thymic radical removal; moreover, eight (20%) patients underwent previous partial pleural resections for pleural relapse. the attention was focalized on the necessity of lung resection, and phrenic muscle resection. The number of resected nodes was evaluated during the pathologic analysis. The number of resected nodes was statistically higher in the DNT group. Post-Operative Outcomes Post-operative complications were reported in 16 patients (40%). Nonspecific HITHOC complications included infectious pneumopathy (n = 6; 15%), prolonged air leak in two cases (5%) and two cases of pyothorax (5%). Specific HITHOC complications were reported as follow: four cases of (10%) acute renal insufficiency without extra renal epuration with spontaneous recuperation after rehydration. One patient experienced spontaneously reversible bone marrow insufficiency. Specific chemotherapy complications were no longer observed due to the natrium thiosulfate nephro-protection protocol. A significant statistical difference was observed between thiosulfate-vs. thiosulfate+ groups in terms of maximal plasmatic peak within the 7-day post-operative period, shown in Figure 1. * the attention was focalized on the necessity of lung resection, and phrenic muscle resectio number of resected nodes was evaluated during the pathologic analysis. The number of r nodes was statistically higher in the DNT group. Post-Operative Outcomes Post-operative complications were reported in 16 patients (40%). Nons HITHOC complications included infectious pneumopathy (n = 6; 15%), prolonged a in two cases (5%) and two cases of pyothorax (5%). Specific HITHOC complication reported as follow: four cases of (10%) acute renal insufficiency without extra epuration with spontaneous recuperation after rehydration. One patient exper spontaneously reversible bone marrow insufficiency. Specific chemoth complications were no longer observed due to the natrium thiosulfate nephro-prot protocol. A significant statistical difference was observed between thiosulfat thiosulfate+ groups in terms of maximal plasmatic peak within the 7-day post-ope period, shown in Figure 1. The median length of stay in the intensive care unit (ICU) and in the hospital were respectively: 1 day, ranged 1-26 and 10 days, ranged 6-36 days. The median length of chest tube removal was 5 days, ranged 4-32. In-hospital mortality was 2.5%. One patient died at day 7 due to a multi-organ failure induced by septic post-operative pleuro-pneumonia. One patient who underwent an extrapleural pneumonectomy died at 3 months due to multi-complicated bronchial fistulae; another patient one died at one year due to septic pneumonia. Long-Term Outcomes The median DFI after HITHOC was 70 months, with a significant difference between DNT and DR, being 49 vs. 85 months (*; p = 0.029), respectively. Eighteen patients (20%) presented subsequent disease recurrence. The most encountered sites of relapse were homolateral pleura in ten patients (25%), contralateral pleura in two patients (5%), and mediastinum in two patients (5%). One patient presented a liver metastasis (2.5%) and another in the axillary lymph node (2.5%); one patient presented lung metastasis (2.5%) and another presented multiple systemic relapses including the pericardium, lung, and spine. More details on the DFI of homolateral pleural vs. contralateral or systemic relapse is shown in Table 3. The median length of survival after HITHOC was 118 months, without significant differences between DNT vs. DR. The median lengths of OS were respectively 94 vs. 118 months. Fifteen patients died during the entire study period, four (26%) due to septic shock induced by pneumonia, four (26%) due to disease progression, two (13.3%) from acute MG crisis, two (13.3%) due to heart failure induced by anthracyclin chemotherapy, one (6.6%) due to a paraneoplastic nephrotic syndrome, one due to COVID-19-induced ARDS, and one from an unknown cause. Survival data are shown in Figure 2, Table 3. We analyzed the effect of the number of resected pleural nodules based on an artificially created threshold of 10. DFI was significantly different in patients with more than 10 resected nodules; medians were respectively 49 vs. 70 months; p = 0.043. This difference did not impact OS, with a median of 118 months in <10 pleural nodules group; median was not found for >10 Pleural nodules group. another in the axillary lymph node (2.5%); one patient presented lung metastasis and another presented multiple systemic relapses including the pericardium, lu spine. More details on the DFI of homolateral pleural vs. contralateral or systemic is shown in Table 3. The median length of survival after HITHOC was 118 months, without sig differences between DNT vs. DR. The median lengths of OS were respectively 94 months. Fifteen patients died during the entire study period, four (26%) due to shock induced by pneumonia, four (26%) due to disease progression, two (13.3% acute MG crisis, two (13.3%) due to heart failure induced by anthracyclin chemot one (6.6%) due to a paraneoplastic nephrotic syndrome, one due to COVID-19-i ARDS, and one from an unknown cause. Survival data are shown in Figure 2, Tab We analyzed the effect of the number of resected pleural nodules based artificially created threshold of 10. DFI was significantly different in patients wit than 10 resected nodules; medians were respectively 49 vs. 70 months; p = 0.04 difference did not impact OS, with a median of 118 months in <10 pleural nodules median was not found for >10 Pleural nodules group. Discussion Complete surgical resection is the cornerstone of the treatment of thymom IVa with pleural and/or pericardial spread is a metastatic disease for which c resection has been deemed unrealistic and intrinsically at least R1 or R2. Fa difficulties in treatment of a metastatic but localized disease with slow ev HITHOC was introduced as a better alternative to extra-pleural pneumonectomy of morbidity, mortality, and overall survival. It also delayed the need for therapies that were used when treatment had limited efficacy in advanced disea In this study, we show that HITHOC is a safe surgical procedure in a multidisciplinary team, with an acceptable 30-day mortality rate of 2.5% and mortality of 5%, in light of the historical results of EPP (respectively 17 and 30% Meanwhile, the morbidity rate is mostly related to post-operative comp interestingly, we were able to reduce the risk of acute renal failure by 10%. The thiosulfate protocol demonstrates a positive influence on renal function, and sho standard for the HITHOC procedure. We were able to increase cisplatine dosag mg/m 2 without adverse effect, and may encourage increasing the dosage to mg/m 2 . However, the optimal regimen for intrathoracic chemotherapy is sti question. The cisplatin regimen has been the most evaluated with positive results penetration, but there is a lack of rationale for a second drug [19,20]. In the prese we used mitomycin, which has replaced doxorubicin for being more eff Discussion Complete surgical resection is the cornerstone of the treatment of thymomas. Stage IVa with pleural and/or pericardial spread is a metastatic disease for which complete resection has been deemed unrealistic and intrinsically at least R 1 or R 2 . Facing the difficulties in treatment of a metastatic but localized disease with slow evolution, HITHOC was introduced as a better alternative to extra-pleural pneumonectomy in terms of morbidity, mortality, and overall survival. It also delayed the need for systemic therapies that were used when treatment had limited efficacy in advanced disease [6,15]. In this study, we show that HITHOC is a safe surgical procedure in a trained multidisciplinary team, with an acceptable 30-day mortality rate of 2.5% and 90-day mortality of 5%, in light of the historical results of EPP (respectively 17 and 30% [16][17][18]. Meanwhile, the morbidity rate is mostly related to post-operative complications; interestingly, we were able to reduce the risk of acute renal failure by 10%. The natrium thiosulfate protocol demonstrates a positive influence on renal function, and should be a standard for the HITHOC procedure. We were able to increase cisplatine dosage to 100 mg/m 2 without adverse effect, and may encourage increasing the dosage to 150-200 mg/m 2 . However, the optimal regimen for intrathoracic chemotherapy is still under question. The cisplatin regimen has been the most evaluated with positive results on tissue penetration, but there is a lack of rationale for a second drug [19,20]. In the present series, we used mitomycin, which has replaced doxorubicin for being more efficient in thymomas [21]. Unfortunately, given the rarity of TETs with pleural spread, a clinical evaluation of different chemotherapy regimens is challenging. In our opinion, a preliminary in vitro model using tumoral thymic epithelial cells should be used as a preliminary evaluation for other potential drugs [22]. DNT should not be a contra-indication and must encourage multi-modal treatment. This involves neo-adjuvant chemotherapy to reduce the volume of the thymic tumor, if not resectable (as recommended in the RYTHMIC guidelines) and a two-step surgery: (1) Radical thymectomy and, one or two months later, (2) Subtotal pleurectomy and HITHOC. The choice of a two-step surgery was guided by two compelling reasons: (1) The necessity of a (2) The use of a postero-lateral approach for HITHOC with a surgery duration of 6-7 h in the operating theatre. Nevertheless, DNT appears ro be a more aggressive disease with a higher rate of B2-B3 thymomas and a higher rate of pleural nodules adversely influencing DFI compared with DT. We report that the number of metastatic pleural nodules influence DFI without impact on OS. The presence of >10 pleural nodules was a significant factor of lower DFI and affected the extent of pleural invasion. Low et al. and Qayyum et al. [23,24] advocated for a MRI with higher sensitivity to predict the peritoneal carcinomatosis index (PCI) before the cytoreductive peritoneal procedure. Standard pre-operative surfaced MRI and PCI adapted to pleura would be helpful for the decision and prognosis of pleural cytoreductive HITHOC procedures. Moreover, in our experience, the number of metastatic pleural nodules encountered during surgery is always more important than those revealed by CT or 18-FDG scan and is, in our opinion, a major reason to prefer subtotal pleurectomy to tumorectomy in an R 1 surgery. One of the big questions currently unanswered is the veracity of HITHOC's procedure. Several response elements can be added. Firstly, in the context of thymoma, it has been clearly demonstrated that the most complete surgical resection is a first-order predictive factor. Secondly, we were able to demonstrate using the French database RYTHMIC that surgical excision of pleural nodules (including HITHOC cases) in stage IVa compared to systemic treatments reduced the risk of recurrence by 60% (HR = 0.4, 95 CI (0.25-062); p < 0.0001) [25]. These results were confirmed by the study of Aprile et al. [26], who were able to demonstrate in a retrospective study that the HITHOC procedure coupled with a cytoreductive pleurectomy gave better results in DFI compared with cytoreduction alone: 88 ± 15 vs. 57 ± 19 months (p = 0.046). Moreover, our results are fully comparable with those previously published showing a 5-year survival rate between 70-100% and a DFI of 61-87% [27]. However, it remains difficult to interpret DFI alternatively published in the form of percent or months; in our opinion, DFI expression in months seems to be more informative for the oncologist. Finally, in the context of this disease, it remains very difficult to set up a randomized comparative study, which could answer these questions with greater certainty. The evaluation of the HITHOC procedure in thymomas remains difficult due to the rarity of the disease. This leads to relative ignorance of standard of care and innovations by physicians. There is great heterogeneity in stage Iva thymoma, in terms of pleural lesion number, size, location, and time interval after initial treatment or diagnosis, which also hampers such assessments. We insist on the major role played by dedicated multidisciplinary networks such as RYTHMIC, ITMIG, ESTS, JART, and CHART to collect data and provide international standards of care. In our opinion, the evaluation of the real impact of HITHOC procedures should regroup all expert centers to define a multicentric study protocol. Evaluation should compare subtotal pleurectomy alone vs. subtotal pleurectomy with HITHOC vs. systemic therapies. The ultimate goal should be to define, in the era of personalized medicine, the best therapeutic option in selected patients to chronicize a metastatic disease with a good quality of life and achieve prolonged outcomes. Conclusions Subtotal parietal and phrenic pleurectomy combined with HITHOC seems to be a valuable option in Masaoka IVa thymomas in highly selected patients. We highlighted a median DFI of 49 and 85 in DNT and DR and OS of 94 and 118 months, respectively. This result offers the possibilities of a long therapeutic time-out in a metastatic disease with a high risk of relapse. In our opinion, such cases must be discussed in dedicated multidisciplinary boards to improve and clarify result of HITHOC. The gold standard would be a randomized trial, which cannot be realized, in an orphan disease with indolent progression. International collaboration would be helpful to organize a prospective database in this setting.	2022-10-19T15:31:29.288Z	2022-10-01T00:00:00.000	{ "year": 2022, "sha1": "72ba42691d653c4015fc05bd6a09651d47b4b9cf", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2072-6694/14/20/5035/pdf?version=1665743208", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "b3e9739bbde25f3cf3217ba9286e0dfa36d24fa0", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
246695501	pes2o/s2orc	v3-fos-license	Chirality at nanoscale for bioscience In the rapidly expanding fields of nanoscience and nanotechnology, there is considerable interest in chiral nanomaterials, which are endowed with unusually strong circular dichroism. In this review, we summarize the principles of organization underlying chiral nanomaterials and generalize the recent advances in the main strategies used to fabricate these nanoparticles for bioscience applications. The creation of chirality from nanoscale building blocks has been investigated both experimentally and theoretically, and the tunability of chirality using external fields, such as light and magnetic fields, has allowed the optical activity of these materials to be controlled and their properties understood. Therefore, the specific recognition and potential applications of chiral materials in bioscience are discussed. The effects of the chirality of nanostructures on biological systems have been exploited to sense and cut molecules, for therapeutic applications, and so on. In the final part of this review, we examine the future perspectives for chiral nanomaterials in bioscience and the challenges posed by them. Introduction Chirality is a property whereby a chiral object is not superimposable upon its mirror image. A substance and its nonsuperimposable mirror counterpart are called 'enantiomers', and have similar physical properties but cause linearly polarized (LP) light to rotate in different directions. Dextrorotatory (D) enantiomers rotate LP light in a right-handed or clockwise direction, whereas the levorotatory (L) enantiomers rotate LP light in a le-handed or counterclockwise direction. This phenomenon is called 'optical activity', and can be described by the conventional spectroscopic technique of circular dichroism (CD), which measures the different absorption of le and right circularly polarized light (CPL) by chiral materials. The anisotropy factor (g-factor) is also used to quantitatively describe the optical properties of chiral materials. Chirality is ubiquitous in nature, occurring in molecules, proteins, nucleic acids, and biological morphologies. 1 The preferences of many biomolecules for specic enantiomers are universal in living bodies. One enantiomer of a chiral material may be a building block for a life activity, whereas the other one is ineffective or even toxic. Chirality is also a fundamental property of molecular recognition during metabolism. The exploitation of chirality at the nanoscale is regarded as one of the most promising areas in nanomaterial research today. The associations between the morphology, scale, charge, and chirality of biomolecules and those of chiral nanomaterials are currently drawing considerable attention in various research elds. Unlike achiral materials, chiral nanomaterials have excellent optical properties, which allow them to be used as a nanotechnological tool for the detection of biological compound. 2 The dissymmetric arrangement of the components of chiral materials, together with the inuence of various extraneous elds, play a crucial role in controlling their properties. 3 In achiral structures, the arrangements of these nanoscale building blocks lack 'handedness' features, limiting their utility in recognition in biosystems. Interestingly, chiral structures can be prepared spontaneously with various approaches, according to human design. The unique spatial arrangements of chiral nanomaterials allow greater control of the collective interactions between the different building blocks of the materials and their interactions with other molecules. Understanding the construction of these nanoscale building blocks into unique chiral spatial architectures allows the orientations and chiral optical properties of these architectures to be controlled. Furthermore, the simulation of the chiral morphologies of nonliving systems, assembled from inorganic metal nanoparticles (NPs), should facilitate the analysis of natural structures in biology. The excellent characteristics of chiral nanostructures mean that they have been extensively studied and have practical value in the elds of modern nanophotonics, including quantum electrodynamics, ultrasensitive biological detection, antibacterial processes, precise chemical analyses, and understanding the chiral effects in living systems. [4][5][6][7] Several approaches have been developed by scientists around the world to precisely prepare chiral nanomaterials. These can be classied into three types: chiral assemblies of achiral building blocks; NPs with intrinsic chiral morphologies; and achiral NPs covered with 2. Fabrication of chiral nanomaterials and their applications in bioscience 2.1 Template-assisted chiral nanomaterials A variety of strategies are used to fabricate chiral nanostructures. 15 The simplest route is based on a chiral template, in which the translation of the conformation of the template into building blocks allows the researcher to guide the self-assembly or synthesis of template-based nanoscale materials. A patterned physical template, such as a gel, micelle, polymer, mesoporous silica, or supramolecular ber, provides a rigid geometric structure. Aer its assembly, the building blocks are arranged around or along the well-dened shape or structure of the template. Luis M. Liz-Marzan's group reported the use of amyloid brils or chiral bers as scaffolds for nanorod (NR) assembly. 16,17 With this method, the dispersion of heterogeneous NRs was guided into a chiral conformation based on the template, and a strong surface plasmon-mediated CD (SP-CD) signal was rst obtained with nonspherical metal NPs. Chiral amyloid was used as the ber backbone template upon which the helical morphology of the NRs was constructed. The assembled NRs showed strong plasmonic CD intensity and a chiral arrangement in solution aer the addition of the bers ( Fig. 2A-D). The signal of the assembly in the plasmon resonance region had low background interference and interacted specically with helical protein brils. The method developed was used to identify these brils in human brain homogenates from patients with Parkinson's disease, conrming its potential application in the detection of clinical neurodegenerative diseases. As well as the nano-assembly of achiral NPs, single NPs with chiral congurations can also be constructed with template mediation. Chiral micelles of 1,1 0 -bi(2-naphthol) (BINOL), a cosurfactant, have recently been used as a template for the seeded growth of anisotropic gold nanocrystals. The adsorption of the micelles to gold NRs induced the assembly of a surfactant (cetyltrimethylammonium chloride [CTAC]) into chiral molecules that formed quasihelical patterns ( Fig. 2E-G). These directed the diffusion of the gold-containing micellar aggregates and induced the formation of wrinkles on the NR surfaces ( Fig. 2H and I). Notably, the optical activity of the NRs could be modulated by varying their aspect ratio, and high anisotropy factors of about 0.20 were obtained from 500 to >1350 nm, which are the highest values reported for colloidal plasmonic NPs in the biological activity window of the nearinfrared region. 18 A theoretical analysis revealed that the chirality originated from the coiling morphology in opposite directions. DNA origami is a versatile and robust template for the construction of chiral nanomaterials. [19][20][21][22] The DNA structures can be tailored to have well-dened spatial congurations, from which plasmonic chiral structures can be constructed through self-assembly. 23 DNA can be used as the construction templates upon which plasmonic NPs are organized into chiral conformations through the ingenious design of DNA origami. 24 The different handedness in space and plasmon coupling can be programmably controlled in highly ordered ways. The research group of Na Liu and Tim Liedl rst designed a recongurable DNA origami template that produced a chiral arrangement of gold NRs with a strong CD signal. 25 The chirality of these assemblies arises from both the structural chirality and plasmonic chirality. Moreover, in different forms of chiral micelles or gels, the handedness or the morphology of the template (DNA origami) can be manipulated as desired, to provide innite potential applications. Therefore, the optical response obtained can generate the ideal symmetrical spectrum. The interaction between light and matter at the nanoscale can also be exploited by walking NRs stepwise, both directionally and progressively, on DNA origami, using DNA as the fuel. 26 With this method, different plasmonic stereoisomers with various chiral centers have been constructed. 27,28 Importantly, the strong CD signal was used to sensitively and selectively detect a specic target sequence, with which the cross angle of the assembly could be immobilized. Dening the state of the plasmonic conguration allowed a distinct optical response to be generated, with great potential utility in biodetection. 29 Tim Liedl's group detected and quantied an RNA sequence of Hepatitis C virus (HCV), with a limit of detection (LOD) of 100 pM, which paved the way to use DNA origami and chiral nanomaterials as rapid detection reagents for pathogenic nucleic acids. 30 Other external inputs was integrated with two kinds of aptamers and a logical gate can be designed based on this platform for biological applications (Fig. 3A-D). 31 It was functionalized with both adenosine triphosphate (ATP) and cocaine aptamers, which can both be activated thermally. By exploiting the high specicity and selectivity of aptamers, more-dynamic plasmonic probes for various molecular targets can be constructed with DNA origami. 27,32-35 So templates other than DNA molecules, such as peptides, have also been used in the assembly of helically arranged gold NPs or nanowires (Fig. 3E). 36,37 A family of helical gold NP assemblies was prepared with C 18 -(PEP M-ox Au ) 2 (PEP M-ox Au ¼ AYSS-GAPPM ox PPF) by Professor Nathaniel L. Rosi's group. [38][39][40][41] The peptide-conjugated molecules contained an organic tail that directed their self-assembly and bound to gold NPs, which displayed strong plasmonic chiroptical activity, with g-factor values of up to 0.04. The morphology and chiroptical activity of the nanowire could be tuned effectively by changing the aliphatic tail length, the helical pitch and size, the shape, and the aspect ratio of the NPs. The positions of methionine and methionine sulfoxide within the peptide sequence could also be changed to control the size and aspect ratio of the NPs, potentially enhancing the chiroptical properties of the helices. The biomolecule templates-based way for the construction of chiral nanomaterials displayed higher biocompatibility than other methods, with consequent potential utility in living systems. Chiral nano-assembly Like the template-assisted methods, self-assembly is one of the commonly used approaches to the fabrication of chiral systems. In this section, we focus on the self-assembly processes that are mediated by small ligands attached on the surfaces of building blocks. Under different driving forces, isotropic NPs are directed into a series of discrete uniform patterns through selfassembly, which by denition, proceeds without human intervention. 42,43 Professor Kotov's laboratory reported the fabrication by self-assembly of chiral inorganic particles with uniform enantiomeric congurations. They transferred nanoscale chirality to complex structures, driven by CPL or chiral amino acids. Strong chiroptical activity was obtained by tuning the noble metal thiolates and surface ligands. The chiral nanoribbons adopted twisted orientations of different handedness aer racemic CdTe NPs were illuminated with CPL, resulting in an enantiomeric excess (e.e.) of 30%. This conrmed that CPL can be used as 'a template' for the self-assembly of enantioselective chiral NPs. 44 Chiral self-sorting was also observed in cysteine (Cys)-stabilized CdTe NPs in methanol, and was attributed to a thermodynamic preference for homochiral NP assemblies. The e.e. of the nal helical shape of the assembled structure was $98%, and the g-factor was 0.01, more than two orders of magnitude higher than that of the NPs. 45 To further modulate chiral optical activities at the nanoscale in real time, chiromagnetic Co 3 O 4 NPs with chirally distorted crystal lattices were synthesized. 46 The chiroptical effects of the paramagnetic NPs in both dispersions and gels were 10 times stronger than those of nonparamagnetic NPs. The transparency of the Co 3 O 4 gels to CPL was tunable under a magnetic eld. The chiralityinduced assembly of nanostructures has also been investigated. The use of noble-metal thiolates of Au-Cys allowed the formation of hierarchically organized particles (HOPs), in which the complexity displayed was extraordinarily higher than that in their biological counterparts or other complex NPs (Fig. 4A). 47 The assembly pathways were dependent on the symmetry of the NPs rather than on their size, which indicated chiralitydependent assembly restrictions. The chiral conguration of the HOPs originated from the staggered arrangement of twisted nanoribbons. Moreover, the chiroptical properties of the HOPs could be tuned by doping them with coinage metals. Based on this report, the chirality of NPs should be considered to play a fundamental role in assembly pathways and warrants further research. By mimicking the long-range organization of chiral molecules in liquid crystals, high optical asymmetry at the nanoscale level was achieved with the assembly of gold NRs into long helical chain-like NRs with an end-to-end orientation aer they were conjugated to human islet amyloid polypeptides (hIAPPs) ( Fig. 4B). Aer they were co-assembled with the nanohelix peptides, the NRs assembled into a long-range chiral form, and the optical asymmetry g-factor of the assembly was more than 4600 times higher than those of the monomer. The chiral response could also be regulated by adjusting the NR size and the helix pitch ( Fig. 4C and D). Based on this design, more chiral nanomaterials with high optical asymmetry can be produced. More importantly, rapid sensing and drug discovery protocols can be developed for complex biological environments based on the plasmonic CD spectrum or polarization rotation imaging of chiral assembly, which has low background interference. 48,49 2.3 Chiral quantum dots (QDs) Chiral QDs are a promising material for applications in biological sensing, photonics, and luminescent devices because they have readily tunable optical activities. 50 Most chiral optically active QDs with an achiral core are prepared with chiral ligands. 51,52 Professor Milan Balaz has studied the synthesis of chiral semiconductor NPs with different chiral ligands, and the relationships between the NP structures and ligands and their optical activity (Fig. 5A). Chiral CdSe QDs with L-or D-Cys ligands were prepared with a postsynthetic ligand exchange method. 53 The optical activity induced showed size-dependent variations, and the origin of the signal was shown to be the orbital hybridization of the highest occupied CdSe molecule with a chiral ligand. Based on these results, further studies of ligand-induced chirality in QDs were undertaken with a thiolfree chiral carboxylic acid (malic or tartaric acid) as the capping ligand for the chiral CdSe QDs (Fig. 5B). 54 Three oxygen-donor groups were shown to be necessary for the induction of chirality in CdSe, which in turn determined the CD signal. Stereoselective porphyrin-driven supramolecular nanoassemblies were also prepared on DNA templates. The photocatalytic activity of these nanostructures and the highly sensitive and selective spectroscopic detection were also studied. Experimental and theoretical calculations demonstrated that the attachment of chiral ligands to the surfaces of QDs generated the induced chirality through a bidentate interaction. To further enhance the g-factors of chiral CdSe QDs, Vivian E. Ferry's group demonstrated that carboxylate-capped CdSe QDs showed more-intense CD signals than thiolate-modied CdSe QDs, with values reaching 7.0 Â 10 À4 (Fig. 5C). 55 The origin of this chirality was examined from the splitting of the exciton by the interaction between the ligand and the QDs. The stereocenter of the ligand correlated positively with the g-factor of the chiral QDs. With these constructive advances, they conrmed that the chiroptical response of QDs can be tailored by the chemical structure of the chiral ligand. The hybridization of chiral ligands with the QD valence bands played a crucial role in their chirality, which was then inuenced by the functional group or binding mode. During the synthesis of QDs, a core-shell structure is one possible interesting design, which enhances the quantum yield of the photoluminescence. Based on this phenomenon, Professor Yurii K Gun'ko examined the synthesis of chiral coreshell QDs and the corresponding chiroptical activity (Fig. 5D). A series of CdS shells, with thicknesses ranging from 0.5 to 2 nm, were coated onto CdSe QDs. 56,57 Aer ligand exchange with chiral Cys, chiral CdSe/CdS core-shell QDs were obtained. The chiroptical response was found to be reduced, whereas the photoluminescence of the QDs was enhanced, as the thickness of the CdS shell increased. These studies provided the foundations for the design of chiral uorescent probes with QDs. The chiral origin and optical properties of chiral semiconductor nanomaterials have been investigated by Professor Zhiyong Tang and colleagues, who have also explored the application of chiral materials to asymmetric catalysis, enantioselective separation, and chiral recognition. To explain the origins of and the changes in the CD responses in chiral semiconductor nanocrystals, in both the ultraviolet (UV) and visible regions, they fabricated chiral cysteine-moleculestabilized CdSe quantum rods (QRs), with greatly enhanced CD responses. As the geometric aspect ratio of the QRs increased, the CD signal improved, and the optical activity of the colloidal semiconductor nanocrystals was induced by chiral organic ligands. The theory was explored with a nondegenerate coupled oscillator model, based on the different polarization of the excitonic transitions. Increasing the asymmetry of the QRs also increased the transition of the linearly polarized excitons, which caused stronger optical activity. 58,59 Intrinsic chirality is also present in cinnabar mercury sulde (HgS) nanocrystals, which have a clearly chiral geometric morphology. [60][61][62][63] This chirality originates in the interplay between the chiral crystallographic lattice and the geometric handedness. Chiral nanolms As well as single chiral NPs or chiral assemblies, twodimensional chiral materials with distinct optical activities can also be produced. Zhiyong Tang's group fabricated largearea long-range ordered ultrathin chiral lms using a bottomup assembly method. Aligned one-dimensional ultrathin gold (Au) nanowires were prepared with the Langmuir-Schaeffer technique, and the subsequent layers were then rotated clockwise or anticlockwise at a predetermined angle to generate leor right-handed ultrathin chiral lms (Fig. 6A). 64 The highest anisotropic factor (0.285) was obtained with three layers of Au nanowire (with an interlayer angle of 45 ) in the wide 300-2000 nm wavelength range. Calculations demonstrated that the origin of the strong optical activity was the helical stacking of the layers of the anisotropic Au nanowire assembly. The universality of the method allowed its application to W 18 O 49 and NiMoO 4 $xH 2 O. This technique was also used to fabricate le-or right-handed photonic crystals, and circularly polarized color reection was observed with the clockwise or anticlockwise horizontal rotation of each layer. 65,66 Kadodwala's group reported the generation of superchiral electromagnetic elds with gold gammadions of planar chiral metamaterials (PCM), which greatly enhanced the sensitivity of chiral molecules in chiroptical detection ( Fig. 6B and C). The optical properties of le-or right-handed panels were used to determine the differences in chiral samples under le-and right-CPL. [67][68][69][70][71] The interaction of chiral molecules with an electromagnetic eld was enhanced by coupling them to the localized surface plasmon resonances (LSPRs) of gold gammadions. The shis in the resonance wavelength induced by the adsorption of different chiral targets were detected for myoglobin, hemoglobin, and bovine serum albumin (BSA), which have high levels of a-helical secondary structure, and for b-lactoglobulin, outer membrane protein A, and concanavalin. Based on this method, the largest dissymmetries in the shis of the LSPRs on adsorption of chiral layers was observed for biomolecules with many b-sheets, and was 10 6 times higher than other optical polarimetry measurements. Professor Shunai Che's group proposed a novel method of template-assisted hierarchical self-assembly for constructing highly ordered chiral mesoporous materials, in which a stack of chiral anionic organic templates induced chiral imprinting and amplication. 72 They also generated mesoscopic structural materials via a chiral-molecule-induced pathway. Multilevel three-dimensional (3D) lms (ZnO, silver, CuO, NiO, and so on), with hierarchical chirality from the atomic to the micron scales, were constructed using a symmetry-breaking agent, and their cooperative assembly effect was studied. 73-76 Each lm contained three levels of chirality, which were identied with structural characterization: (1) primary distorted nanoakes/ nanoplates with atomic crystal lattices; (2) secondary helical stacking of these nanoakes to form nanoplates; and (3) tertiary micrometer-sized circinate aggregates of the chirally arranged nanoplates. Lattice defects were induced by the chiral center of the amino acid, which ultimately led to a circinate helical conguration. This research group suggested that the distorted crystalline structure and the helically aggregated nanoplates in these nanocrystals were smaller than Bohr's radius. The dynamic coulomb interaction in a dissymmetric eld caused electron-transition-based optical activity. Moreover, scatteringbased optical activity, CPL, and Raman optical activity were also generated with these nanocrystals, which could be applied to the chiral selection of amino acids. Very recently, arrays of chiral nanostructured Au lms with high anisotropy were prepared for the discrimination of enantiomers with surfaceenhanced Raman scattering (SERS) under linearly polarized or unpolarized light. They showed great versatility, detecting almost all enantiomers. 77 Plasmonic chirality at the nanoscale Plasmonic nanostructures are one type of building block that can be used for chiral nanoprobe construction because they have enhanced electromagnetic near-elds. This method has received much attention because the nanostructures are easily accessible under mild synthesis conditions. As we mentioned above, chiral plasmonic nanostructures can be obtained with template-assisted or ligand-induced methods. Their intrinsic chiral morphology provides clear structural chirality. 78 Moreover, like the structural chirality of NPs, the plasmon-coupled effect generates a strong plasmonic CD signal. 79 A plasmonic NPs can induce the chirality of ligand amplication when it is closely located to the hotspot region, which is reportedly a promising approach for ultrasensitive biodetection. 80,81 However, structural chirality and plasmon chirality are hard to distinguish in ensemble analyses. Professor Stephan Link investigated the origin of the optical activity of chiral molecules and nanomaterials, and identied the physical principles governing the interactions between plasmonic NPs and between these NPs and their chiral environments, 82 with the goal of using plasmons, or light photons, to probe materials and initiate chemical reactions. Single-particle circular differential scattering spectroscopy (CDS) was used to differentiate the origin of chirality in the structural chirality of plasmonic aggregates or in the plasmon-coupled CD signal from that in gold NR-BSA complexes (Fig. 7A-D). The aggregation induced by NR-BSA produced a CDS signal, whereas no detectable signal was observed from single NR-BSA complexes or even single NRs with a BSA monolayer. The strong CDS signal from the aggregates was shown to originate from the intrinsic structural chirality, with a right-handed conguration. Electromagnetic hotspots enhanced the BSA-induced plasmon-coupled CD signal, which was conrmed with Raman spectroscopy. The conformation of the NR-BSA aggregates was visualized systematically using high-angle annular dark-eld scanning transmission electron microscopy (HAADF-STEM), with tomographic reconstruction from different tilt angles and nitedifference time-domain (FDTD) simulations. The researchers found that the nonparallel dimer was chiral, with a strong CDS signal at 720 nm. Enhanced plasmon-coupled CD was also observed. The enhanced plasmon-coupled CD of a parallel achiral dimer was explained by the positions of the BSA molecules in the gaps between the NRs. The NR aggregates induced with NaCl, as a control, showed only structural chirality. From this study, the origin of chirality from complexes system can be determined from CDS measurements. Dangyuan Lei's group fabricated a plasmonic chiral nanomaterial and investigated the chiral coupling through photonic spin-orbit interactions. 83 With achiral nanostructures, they found that chiral cysteine molecules induced chiral optical activity on the surfaces of the NPs by inducing the formation of hydrogen-bond-connected helical networks of chemisorbed molecules on the NPs. 84 This strong coulomb coupling and the local electric eld of the plasmonic nanomaterial further enhanced the CD resonance. Notably, its intensity could be precisely dened and dynamically tuned by adjusting the solution temperature, pH, or the ions present. The detailed bio-inspired chiral geometries of metallic nanostructures have been investigated by Ki Tae Nam's group. 85,86 The chiral geometries of nanostructures are considered the key factors in enhancing their chiroptic responses (Fig. 7E). 87 The biomolecule-based synthesis of 3D chiral plasmonic materials was examined in terms of peptide selfassembly and peptide interfaces with metallic materials. By investigating the mechanisms by which chiral geometries are involved in the formation of metallic nanostructures, these researchers used peptide-sequence-specic interactions with high-index surfaces to control the growth and optical properties of nanomaterials. A series of gold NPs with low-Miller-index surfaces were synthesized with a seed-mediated method. 88 Aer these NPs had interacted with differently handed cysteine or cysteine-based peptides in the presence of Au + ions, the lowindex-plane exposed gold NPs evolved into high-index-plane NPs. The researchers demonstrated that the morphology and chiral optical properties of the NPs could be modulated by tuning the peptide sequence, amino acids, or gold seed. Because the growth rates of the chiral high-index planes of the NPs differed, the asymmetric evolution of the NPs ultimately led to the formation of helicoid morphologies. Based on this method, the strongest optical activity (with a g-factor of 0.2) was obtained when an octahedral seed was used. Chiral nanocrystals As well as plasmonic nanomaterials, chiral nanocrystals of inorganic materials have been prepared. Gil Markovich's group demonstrated that the interaction between chiral ligands, the spontaneous symmetry in lattice breaking, and the enantioselective nucleation of the growth chemistry during the crystallization process allowed the handedness of the nanocrystals to be controlled. 89,90 Using thiolated chiral biomolecules as efficient reducing agents, chiral Te or Se NP precursors could be obtained. The handedness, chiral optical activity, and shape of the NPs could also be determined by the biomolecule used. The researchers found that the chiral conguration of the nanocrystals was formed by a unique self-assembly process, which was conrmed with a discrete dipole approximation simulation. The chirality originated not only from the chiral atomic lattice, but also from the mesoscale shape, in which the twisted ridges crossing diagonally between vertices dened the handedness. Although chiral additives can determine a chiral shape, whether it is a necessary condition for a chiral crystal structure had not been tested. Therefore, Professor A. Paul Alivisatos's group investigated, in detail, the growth of chiral tellurium (Te) nanocrystals with either chiral or achiral ligands (Fig. 7F). They observed the chiral arrangement of a distinct structure of at facets of thick trigonal-bipyramidal Te nanocrystals aer chiral penicillamine was added to the reaction. 91 However, the addition of a pure chiral ligand to the nal reaction generated an e.e. (85%) of only one mirror, not 100%, indicating that the chiral ligand affected the nucleation and growth rates of the NPs. More importantly, when the achiral ligand mercaptopropionic acid was used instead of penicillamine, equivalent le-and right-handed nanocrystals were formed, demonstrating that chiral additives are not necessary for the formation of a chiral shape. Further studies with aberration-corrected HAADF-STEM and 4D scanning electron microscopy showed that lattice twisting occurred during the growth process, conrming that screw-dislocation-mediated layer-by-layer growth induced the chiral polyhedrons of the nanocrystals. That study was the rst to demonstrate that screw-dislocation-mediated growth is the origin of the morphological chirality of the facets in single crystals. Chiral biosensor and their biological effects The research group of Xu and Kuang investigated the construction of chiral nanoprobes and their biological effects, [92][93][94] and comprehensively studied the interactions between chiral nanomaterials and biosystems. An ultrasensitive detection method for biomolecules, including biomarkers of disease, biotoxins, bacteria, metal ions, and markers of cell metabolism, was established with self-assembled plasmonic chiral probes, based on their strong optical activity in the visible spectrum (Fig. 8A). [95][96][97][98][99] The enhanced plasmonic CD signals in the visible range and the target-induced specic congurational changes in well-dened sensors resulted in much lower LODs or even allowed the detection of single molecules than other method. Dual signals produced by multifunctional imagingbased nanostructures were used to quantitatively detect cancer markers in both living cells and in vivo, extending the toolbox of chiral materials for the clinical diagnosis and treatment of cancer. 80,100 Chiral NPs have been developed for the specic recognition of biomolecules, and have been shown to selectively cut both DNA and protein. Chiral nanomaterials have many features in common with natural chiral biomolecules. Therefore, the key purpose of our studies was to construct chiral nanostructures that not only mimic the functions of proteins, but also the scale matching with upstream or downstream metabolic pathways. Our initial studies have shown that aer the appropriate chiral ligand (Cys) was identied, truncated tetrahedral chiral CdTe NPs with a diameter of 4.5 AE 0.3 nm cut salmon-sperm DNA under CPL illumination (Fig. 8B). 101 Notably, these NPs sequence-specically targeted the GATATC motif and siteselectively cut between adenine and thymine. The mechanism underlying this specicity of binding originated from the conformational match between the exible targeted sequence and the truncated tetrahedral shape of CdTe. Based on this work, NPs with different chiral morphologies have been designed for use in biomolecule recognition. Recently, chiral Fe x Cu y Se NPs were fabricated with broad chiral optical activities, in the range of 400-1000 nm. 102 These chiral NPs interfered with b-amyloid (Ab) under 808 nm illumination, both in vitro and in vivo (Fig. 8C). The D-type NPs had much greater affinity for Ab42 brils than the L-type NPs, and exerted a better therapeutic effect on neurodegenerative diseases. Copper(I) sulde NPs were reported to recognize and cleave the core antigen of Hepatitis B virus (HBcAg), thus blocking viral transmission in living cells. Further conrmation was provided in a chronically infected transgenic mouse model, in which the therapeutic effect of the chiral NPs was better than that of a clinical drug. This illustrates the potential utility of chiral NPs as antiviral agents. 103 Advances in this direction, clarifying the biological interactions between NPs of different handedness and living systems, have also been made. Chiral nanoassemblies modied with different glutathione (GSH) enantiomers have been used to regulate basic metabolic processes. Chirality-dependent D-GSHcoated structures displayed an enhanced autophagy-inducing ability, which was attributed to the different accumulation state of the structures. The degree of autophagy could also be quantied in situ from the intensity of the CD signal. 104 A CPL-triggered mechanical force has also been identied in chiral assemblies. 105,106 Based on the design of a DNA-based chiral assembly, CPL was used to stimulate the differentiation rate of neural stem cells into neurons (Fig. 8D). The chiral assembly became specically entangled with the cytoskeletal bers of the cells and an asymmetric force was generated under illumination, conrmed with both experimental and theoretical evidence, which deformed the actin in the cytoskeletons of the cells. The differentiated neurons were successfully implanted into the hippocampi of mice with Alzheimer's disease (AD), paving the way to utilizing the biological effects of CPL and chiral nanomaterials for biomedical applications. 107 Notably, the research group of Tang and Liu showed that chiral gold NPs with a size of 3.3 nm had anti-Ab therapeutic potential aer they were coated with chiral GSH. 108 A Monte Carlo simulation was used to show that the proper NP size was required to reduce the aggregation of Ab42 peptide chains. The chiral NPs not only displayed chiral recognition to inhibit the aggregation of the Ab42 peptide, a prominent histopathological marker of Alzheimer's disease, but also crossed the bloodbrain barrier aer their intravenous administration. The D-GSH-coated NPs inhibited Ab42 brillization much more strongly than those coated with L-type GSH, by forming hydrogen bonds and electrostatic interactions with the peptides, both in vitro and in vivo. This study conrmed the potential utility of chiral nanomaterials as nanomedicines, based on their size-and stereo-determined recognition of peptides or proteins. While exploring the chiral interactions between NPs and living systems to eliminate the size effect, the group of Robert Langer and Ana Jaklenec recently reported the chiralitydependent penetration efficiency of Co 3 O 4 supraparticles (SPs) into living cells. 109 In mechanistic studies, D-Cys-modied SPs showed much greater affinity for lipid bilayers than L-Cys-modied SPs, increasing their cellular internalization of the SPs. The stability of the D-SPs in mice was greater and their biological half-lives longer than those of L-SPs because the inert capping was not digested during metabolism. In a similar study of chiral Se NPs by Tianfeng Chen's group, a positron emission tomography (PET)-based method showed directly that L-GSH-modied Se NPs had a broader biodistribution in major organs than D-or DL-GSH-modied Se NPs. The researchers found that the homology between the NPs and the cytomembrane caused greater cellular uptake of the L-NPs, which then scavenged oxygen in living cells. 110 Conclusions and perspectives Although considerable advances have already been made in this area, further efforts are still required. First, a huge gap still exists in precisely dening the congurations of chiral NPs and molecules. Predictable stereoselective interactions between chiral materials and biomolecules will allow the biological processes of life to be controlled. The chiral recognition in metabolism, which is a fundamental process in the living body, is still an enormous challenge. Second, the development of much-more-sensitive detection methods for many biomolecules is still required. Stimulus-responsive chiral probes that can dynamically read variations in the biological environments of targets by themselves are particularly desirable. Chirality-based imaging devices for biosystems will also be extended to clinical applications. 111 Third, the chiral effects of different nanomaterials are highly diverse. Current studies have shown that not only the surface properties, but also the size, scale, type, and even the state of nanomaterials inuence the biological effects of nanostructures with different handedness. Comprehensive and fundamental investigations of the key factors that affect the behaviors of chiral nanomaterials in living cells are essential, and could unravel the mystery of the origin of chirality in life. Further studies of chiral nanomaterials will build important bridges between these materials and living systems. The characteristics of chiral origins have inspired researchers in materials science to create chirality in various ways, even beyond natural architectures. The exploration of chiral structures with extraordinarily high anisotropic factors, beyond those of plasmonic materials, is also warranted. A new and promising route to achieving this has emerged with the regulation of chiral properties by CPL or magnetic elds. [112][113][114] Moreover, the specic recognition of biomolecules by chiral NPs has inspired us to create much more exquisite chiral structures to precisely modulate the performance of biological systems in living cells. Future research will mainly focus on the behaviors of chiral nanomaterials with other basic physiological functions, such as in immunity, metabolism, nutrition, and nerve conduction. The design of biodegradable and biocompatible properties in chiral NPs may further extend their application in the clinical context to chiral sensing, disease treatment, and antiviral defenses, such as against SARS-CoV-2, Human immunodeciency virus, and HBV. In summary, promising methods for the construction of chiral nanostructures with various building blocks have been described. As this eld expands, there is no doubt that new technologies, new mechanisms, and new directions for designing chiral materials will be developed in the near future. Conflicts of interest The authors declare no competing nancial interest.	2022-02-10T16:15:58.603Z	2022-02-08T00:00:00.000	{ "year": 2022, "sha1": "94f5b9d3028797d4ed59cafbb302c8acce1f95b1", "oa_license": "CCBY", "oa_url": "https://pubs.rsc.org/en/content/articlepdf/2022/sc/d1sc06378b", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "f926b048e9fdfd55a162c154108dac8cefb9f4bc", "s2fieldsofstudy": [ "Materials Science", "Biology", "Engineering" ], "extfieldsofstudy": [ "Medicine" ] }
256897026	pes2o/s2orc	v3-fos-license	Control of antiferromagnetic spin axis orientation in bilayer Fe/CuMnAs films Using x-ray magnetic circular and linear dichroism techniques, we demonstrate a collinear exchange coupling between an epitaxial antiferromagnet, tetragonal CuMnAs, and an Fe surface layer. A small uncompensated Mn magnetic moment is observed which is antiparallel to the Fe magnetization. The staggered magnetization of the 5 nm thick CuMnAs layer is rotatable under small magnetic fields, due to the interlayer exchange coupling. This allows us to obtain the x-ray magnetic linear dichroism spectra for different crystalline orientations of CuMnAs in the (001) plane. This is a key parameter for enabling the understanding of domain structures in CuMnAs imaged using x-ray magnetic linear dichroism microscopy techniques. Ab initio calculations indicate that the stable configurations of the staggered magnetization in tetragonal CuMnAs lie in the (001) plane, where a biaxial magnetic anisotropy is expected due to the crystal symmetry 2,16 . However, the tetragonal polytype of CuMnAs is stabilized by growth on III-V substrates (GaP or GaAs), which leads to an in-plane uniaxial magnetic anisotropy 16,17 . Similar anisotropies are commonly found in FM/III-V films, due to the broken symmetry of the III-V surface 18 . Here we present a study of the magnetic coupling and XMLD spectra in a bilayer film consisting of FM Fe and AF CuMnAs. We combine XMLD as well as x-ray magnetic circular dichroism (XMCD) to obtain element specific information on the FM layer as well as both compensated and uncompensated magnetic moments in the AF layer. In crystalline materials, the XMLD in particular contains rich information on the atomic and magnetic structure. Crystalline anisotropy of XMLD spectra, in which the spectral lineshape depends strongly on the direction of the x-ray polarization vector with respect to the crystallographic axes, has been observed in theoretical and experimental studies of a wide variety of magnetic materials including metals 19,20 , oxides 21-23 and to the x-ray helicity vector, and the difference (XMCD) spectra, at sample temperature 250 K. The Mn XMCD is scaled by a factor of 10 for clarity. diluted magnetic semiconductors 24 . Here we utilize the exchange coupling between the Fe layer and rotatable AF CuMnAs spins to reveal the anisotropic XMLD spectra for tetragonal CuMnAs, which are compared to ab initio calculations. Methods and Results Growth, structure and magnetometry. The sample studied consists of a 2 nm Al/2 nm Fe/5 nm CuMnAs film grown on a GaP(001) substrate by molecular beam epitaxy. The substrate temperature during growth was 260 °C for the CuMnAs layer and 0 °C for the Fe layer and the protective Al cap. The layers were grown in the same ultra-high vacuum chamber, to ensure a clean interface between them. Previous studies have shown that tetragonal CuMnAs is lattice-matched to GaP(001) through a 45° rotation of the unit cell 2 . The measurements described below confirm the epitaxial relationship Fe(001) [110] \|\| CuMnAs(001) [100] \|\| GaP(001) [110]. Figure 1b and c show magnetization loops for the film measured by superconducting quantum interference device (SQUID) magnetometry along the in-plane [110] and [100] directions of the GaP substrate, at temperatures of 200 K and 2 K respectively. Negligible exchange bias is observed, which we attribute to the low in-plane anisotropy of the CuMnAs and its subsequent easy coherent rotation. This is supported by the XMLD data in the following sections. The rounded shape of the loop is ascribed to crystalline disorder, due to the large lattice mismatch between Fe and GaP (001). X-ray magnetic circular and linear dichroism measurements. The XMCD and XMLD measurements were performed on beamline I06-1 of Diamond Light Source, using total electron yield detection and a superconducting vector magnet in which magnetic fields can be applied in any direction. XMCD spectra were measured with the x-ray beam at a grazing angle of 25° to the sample surface, and with a magnetic field of 1000 Oe applied along the beam direction, as illustrated in Fig. 1d. Figure 1e and f show the Fe L 2,3 and Mn L 2,3 x-ray absorption and XMCD spectra from the sample, at a temperature of 250 K. The Mn XMCD is very weak and of opposite sign to the Fe XMCD, indicating a small net Mn magnetic moment which is antiferromagnetically coupled to the Fe layer. The antiparallel alignments of the Fe and CuMnAs magnetic moments is in contrast to Fe 1−x Mn x binary alloys, for which the Mn moment is small and parallel to the Fe 25 . The magnitude of the XMCD asymmetry (I + − I − )/(I + + I − ), where I + and I − are the Mn L 3 peak heights above background for photon helicity parallel and antiparallel to the magnetic field, is around 1%. As shown in Fig. 1a, the magnetic structure in CuMnAs consists of FM (001) planes which are AF coupled to the neighbouring sublattice planes. Therefore, the interface plane of CuMnAs may be expected to consist of uncompensated Mn magnetic moments. Due to the finite probing depth of the total electron yield XMCD measurement, the signal from the uncompensated interface layer is not fully cancelled by the opposite oriented layer below it. The XMCD from the AF ordered CuMnAs film will be smaller than for a fully The XMLD spectra were obtained with the x-ray beam at normal incidence, taking the difference between absorption spectra measured with the x-ray linear polarization vector parallel to the [110] and [110] axes of the GaP substrate. A 1000 Oe magnetic field was applied along either the [110] or [110] axes, with a small out-of-plane tilt in order to increase the electron yield signal. It was verified that the small out-of-plane component of the field did not affect the spectra. The experimental geometry is illustrated in Fig. 2a. The XMLD spectra at the Mn L 2,3 and Fe L 2,3 edges at 250 K are shown in Fig. 2b and c respectively. The XMLD spectra are shown as a fraction of the L 3 absorption peak height above background. The Mn L 3 XMLD signal is larger than that of the Fe and comparable to that of a 10 nm CuMnAs single layer 16 . Given the large size of the Mn XMLD signal, it can be inferred that it is due to the compensated antiferromagnetic Mn moments in the CuMnAs film rather than the small number of uncompensated moments at the interface. Most strikingly, the same XMLD signal, but with opposite sign, is observed when the applied magnetic field is applied in the orthogonal direction. The reversal of the XMLD spectrum is expected for the FM layer if the Fe magnetization orients parallel to the magnetic field. The observation of similar behaviour for the Mn XMLD indicates that the staggered magnetic moments in the CuMnAs layer have a uniaxial orientation and are exchange coupled to the Fe layer, following the reorientation of the Fe magnetization under the applied magnetic field. The rotation of the AF spins is also observed at 300 K and 2 K, although the magnitude of the XMLD is slightly reduced compared to its value at 250 K, as shown in the inset to Fig. 2c. The smaller value at 2 K may be due to competition between the interlayer exchange coupling and magnetocrystalline anisotropy in the CuMnAs layer. Figure 2d and e compare XMLD spectra measured for x-ray polarization and applied magnetic fields along the in-plane [100] and [110] crystal axes. For both the Mn and Fe L 3 absorption edges, the sign and lineshape of the XMLD depend strongly on the crystallographic direction. The Fe L 2,3 XMLD spectra shown in Fig. 2d are in good agreement with previous studies of epitaxial Fe films on GaAs(001) 20 . This confirms that the Fe layer is epitaxial with in-plane crystal directions parallel to those of the substrate. Electronic structure calculations. The Mn L 2,3 XMLD spectra shown in Fig. 2e are compared to ab initio calculations shown in Fig. 2f. The theoretical XMLD spectra were obtained from LDA + U electronic structure calculations 2 using the approach of ref. 13, which neglects electronic correlations and core hole effects. The finite core hole lifetime was mimicked by lorentzian broadening of 0.4 eV. The calculations reproduce some of the main features in the experimental spectra, including the relative separations of the main peaks and their reversal in sign between the different crystal orientations. Additional features on the low-energy side of the L 2 and L 3 peaks in the calculated spectra are not observed in the experiment. The experimental XMLD spectra are defined as the absorption spectra for parallel x-ray polarization and applied magnetic field, minus the absorption spectra for perpendicular x-ray polarization and applied magnetic field. Similarly, the calculated XMLD are the absorption for AF moments parallel to x-ray polarization, minus the absorption for AF moments perpendicular to polarization. Taking into account the 45° rotation of the CuMnAs crystal with respect to the GaP substrate 2 , the sign of the main peaks is in agreement between theory and experiment for both crystal orientations. The comparison of the measured spectra to the calculation therefore indicates that the AF spin axis in the CuMnAs layer is aligned collinear with the external magnetic field, i.e., the interlayer exchange coupling favours a collinear alignment of the FM Fe and AF CuMnAs magnetic moments. Discussion From the XMCD and XMLD results described above, we can infer the following. The Mn XMCD is consistent with the interface atomic layer of the CuMnAs film orienting antiparallel to the epitaxial FM Fe layer as well as to Figure 2. Rotation of the staggered AF moments due to exchange coupling, and anisotropic XMLD spectra. (a) Experimental geometry for the XMLD measurements. (b) Fe L 2,3 and (c) Mn L 2,3 XMLD spectra, obtained as the difference between absorption spectra measured with x-ray linear polarization vector along the [110] and [110] directions of the GaP substrate, with applied magnetic field along [110] (thick lines) and along [110] (thin lines). The inset to (c) shows the magnitude of the Mn L 3 XMLD peak as a function of temperature. (d) Fe L 2,3 and (e) Mn L 2,3 anisotropic XMLD spectra, obtained from the difference between absorption spectra with parallel and perpendicular configurations of the x-ray polarization and the 1000 Oe applied magnetic field, for fields along 〈110〉 (thin blue lines) and 〈100〉 (thick red lines) in-plane axes. The experimental XMLD spectra in (b-e) are measured at temperature T = 250 K. (f) Calculated Mn L 2,3 anisotropic XMLD spectra for tetragonal CuMnAs. the neighbouring CuMnAs magnetic sublattice, although other possible contributions to the measured XMCD signal (e.g. bulk uncoupled moments or interfacial alloying) cannot be ruled out. The AF CuMnAs spins have a collinear coupling to the Fe layer. The AF spins in the CuMnAs layer are rotatable by reorienting the Fe magnetization under relatively small external magnetic fields. This is in contrast to for example CoO/Fe epitaxial layers, where the AF spin configuration is largely frozen for thicknesses above ≈3 nm 13 . Interlayer exchange coupling therefore provides a means to rotate the orientation of compensated AF materials, which are hard to manipulate directly using external magnetic fields. For tetragonal CuMnAs, this may be combined with manipulation of the magnetic order using spin-orbit torques 9, 27 , and electrical 9 or magneto-optical 11 detection, for future hybrid FM/ AF spintronic applications. Also significantly, the description of the XMLD lineshape in CuMnAs will allow for greater understanding of the domain structures imaged by XMLD. This is likely to become a field of great interest owing to the potential application of AF CuMnAs.	2023-02-16T16:21:38.280Z	2017-09-11T00:00:00.000	{ "year": 2017, "sha1": "e455549f4b1dddd88da8005db102a6457c9619c2", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41598-017-11653-8.pdf", "oa_status": "GOLD", "pdf_src": "SpringerNature", "pdf_hash": "e455549f4b1dddd88da8005db102a6457c9619c2", "s2fieldsofstudy": [ "Physics", "Materials Science" ], "extfieldsofstudy": [] }
6862339	pes2o/s2orc	v3-fos-license	Stability and excitations of solitons in 2D Bose-Einstein condensates The small oscillations of solitons in 2D Bose-Einstein condensates are investigated by solving the Kadomtsev-Petviashvili equation which is valid when the velocity of the soliton approaches the speed of sound. We show that the soliton is stable and that the lowest excited states obey the same dispersion law as the one of the stable branch of excitations of a 1D gray soliton in a 2D condensate. The role of these states in thermodynamics is discussed. The small oscillations of solitons in 2D Bose-Einstein condensates are investigated by solving the Kadomtsev-Petviashvili equation which is valid when the velocity of the soliton approaches the speed of sound. We show that the soliton is stable and that the lowest excited states obey the same dispersion law as the one of the stable branch of excitations of a 1D gray soliton in a 2D condensate. The role of these states in thermodynamics is discussed. Solitons in Bose-Einstein condensates (BECs) have recently attracted much attention and have been studied both theoretically and experimentally. The Gross-Pitaevskii (GP) equation admits solitonic solutions corresponding to a local density depletion (i.e., gray and dark solitons). They have been created in 3D BECs 1 and their decay into vortex rings has been observed 2 . In 2D, an interesting type of soliton corresponds to a self-propelled vortexantivortex pair that, when it moves at a velocity close to the Bogoliubov sound speed, c, takes the form of a rarefaction pulse 3 . The energy E and momentum P were calculated in Ref. 3 , and it was found that the curve E(P ) is below the Bogoliubov sound line, approaching it in the low momentum limit. We investigate the excitation spectrum. Here we present our first results for solitons moving with velocity close to c, for which one can reduce the GP equation to the simpler Kadomtsev-Petviashvili (KP) equation 3,4 . Let us consider a 2D condensate with a soliton moving at velocity V in the x-direction. If the density at large distances is n ∞ , one can define the healing length ξ =h/(2mgn ∞ ) 1/2 , where g is the mean-field coupling constant and m is the mass of the bosons. One can introduce the dimensionless variables x → ξx, y → ξy and t → mtξ 2 /h, the normalized order parameter Ψ → √ n ∞ Ψ and the velocity is the sound speed. In the frame moving with the soliton, the GP equation The order parameter can be written as Ψ = n 1/2 e iS . In the limit of V → c, one can introduce the small parameter ε ≡ The density depletion associated with the soliton becomes shallow and one can expand the density and the phase as To the lowest order in ε, the GP equation gives ∂ x s = −εf / √ 2 and the function f obeys the KP equation where we have introduced the stretched variablesx = −εx + ε 3 t/(2 √ 2), i) 1D gray soliton in 2D condensate. By assuming the stationary solution of (4) to be independent ofỹ, Eq. (4) reduces to the Korteweg-de Vries (KdV) equation and the solu- , corresponding to a gray soliton. The linear stability of this solution can be studied by taking f (x,ỹ,t) = f 0 (x − 2t) + ψ(x − 2t)e i(kỹ−ωt) and linearizing Eq. (4) in ψ(x). One gets This equation was already solved by Zakharov 5 , by applying the inverse scattering method, and also by Alexander et al. 6 . The excitations which are localized alongx and periodic alongỹ have dispersion lawω . For a given ε, the eigenfrequencies ω in the original units of the GP equation (1) are given by ω = ε 3ω /(4 √ 2), and the wavevector transforms as k = ε 2k / √ 2, so that For k > ε 2 3/8 the frequency is real, while for k < ε 2 3/8 it is imaginary, causing the instability of the soliton against bending. 4 In the limit ε → 0, the region of instability becomes vanishingly small and, for k ≫ ε 2 , the stable branch of excitations becomes ω = (2/27) 1/4 k 3/2 . It is worth noticing that this dispersion exhibits the same power law as for capillary waves on the surface of a liquid. ii) 2D soliton. The solution of Eq. (4) corresponding to a 2D soliton is 7 This function is plotted in the upper left frame of Fig. 1. In the units of the GP equation (1), the density profile of the soliton is found by inserting (7) into (2). Its size along x and y is of the order of 1/ε and 1/ε 2 , respectively. To the first order in ε, the momentum P and the energy E per unit length are 3 As it must be, the soliton has a sound-like dispersion in the limit where KP equation is valid, that is, when P → 0. Now we consider fluctuations of the form f (x,ỹ, t) = f 0 (x − 2t,ỹ) + ψ(x − 2t,ỹ)e −iωt and linearize Eq.(4) by ψ. Thus, ψ(x,ỹ) satisfies Looking for bound excited states (i.e., ψ → 0 at large distances), it is convenient to integrate Eq. (9) twice alongx and introduce the function φ(x,ỹ) such that ψ = ∂ 2 φ ∂x 2 . The equation for φ is We numerically solve this equation in a square box of size L, by expanding φ as φ(x,ỹ) = ν,µ φ νµ χ 1,ν (x)χ 2,µ (ỹ) with χ 1,ν (x) = L −1/2 e i2πνx/L for \|ν\| ≤ l. Concerning χ 2,µ (ỹ) we note that Eq. (9) is invariant forỹ → −ỹ, so that the function φ is either even or odd function ofỹ. If it is even, then one can take χ 2,µ (ỹ) = (2/L) 1/2 cos(2πµỹ/L) for 1 ≤ µ ≤ l and χ 2,0 (ỹ) = (1/L) 1/2 . If it is odd, one can take χ 2,µ (ỹ) = (2/L) 1/2 sin(2πµỹ/L) for 1 ≤ µ ≤ l. One thus obtain the following matrix equation with q x = 2πν/L, q y = 2πµ/L and The size of the matrix is N × N , where N = 2l(l + 1) for φ even, and N = 2l 2 for φ odd. Typical values in our calculations are l = 70 and L = 60. It is found that all the eigenvalues are real. This is consistent with the results of Refs. 8,9 and shows the linear stability of the 2D soliton. One can expect this result also from pure kinematic considerations. An excited state is unstable if it can decay into several phonons. Energy and momentum must be conserved in this process. It was proved by Iordanskii and one of the authors 10 that for a Bogoliubov-like spectrum the conservation laws can be satisfied only if the dispersion law E(P ) is above the sound line. Since the dispersion law of the 2D soliton is below the sound line, the soliton is stable. We find stable localized modes with positive eigenvalues. Examples are shown in Fig. 1 profile of the soliton decays algebraically, ψ(x,ỹ) decays rather slowly. It is found that the real and imaginary part of ψ are either even or odd functions ofx and each eigenvalue is degenerate, corresponding to even and odd functions ofỹ. By looking at the oscillations of the eigenvectors ψ in the soliton region, one can estimate the transverse and axial wavevectors,k x andk y respectively. In Fig. 2 we plot the calculated eigenfrequenciesω as a function ofk y for the lowest bound states (points with error bars, where the error bars are of the order of the inverse of the axial size of the soliton). In the same figure we plot the dispersion law (6) of the stable branch of excitations of a 1D gray soliton. The spectrum is very similar. The 2D soliton have a discrete spectrum of bound states, due to its finite size, and all wavevectors are above the threshold for the instability of the 1D soliton. Notice that the soliton oscillations can be subject to damping by emission of sound waves. This means that the energy levels of Fig. 2 have to be considered as resonance states. 11 However, there are reasons to believe that this damping is small and can be neglected both in the calculation of the energy levels and in the thermodynamic considerations below. We plan to investigate this damping in the future. As it was noticed in Ref. 3 , this 2D soliton can contribute to the specific heat C of the 2D Bose gas. Indeed, if one applied the usual Bose statistics to the soliton branch, due to its sound-like dispersion (8), its contribution would be C ∝ T 2 as for phonons. The presence of the excited states of the soliton can change the situation. The inclusion of these excitations is an open problem; here we just want to provide an argument which illustrates its nontriviality. One can take into account the excitations of solitons by assuming that the solitonic branch of the spectrum depends on two quantum numbers: P and k. For example, within the limits of applicability of the dispersion law (6), one can write E (P, k) = c \|P\| + βk 3/2 . One has to accept the hypothesis that the number of states in the interval d 2 P is, as usual, Sd 2 P/ (2πh) 2 , where S is the sample area. On the contrary, the quantization rule for k is fixed by the length of the soliton in the y direction, R y , which behaves as R y ∝ ξ/ε 2 . Correspondingly, the number of states in the interval dk is ∝ R y dk/(2π) 3 ∝ (ξ/ε 2 )dk. The energy of the gas is thus E ∝ d 2 P dk ε −2 E(P, k)[exp(E(P, k)/T ) − 1] −1 . A simple calculation shows that the relevant values are E ∝ ε ∝ T, P i ∝ T with i = x, y, and k ∝ T 2/3 . This gives the specific heat C ∝ T 2/3 , which exceeds the phonon contribution at low T . Of course, this result cannot be considered as a rigorous one, in particular because solitons occupy an area ∝ 1/ε 3 and they can overlap at low temperature. Nevertheless it shows that the problem of the low-temperature specific heat of the 2D Bose gas deserves both theoretical and experimental investigation. A direct diagonalization of the Hamiltonian or Monte Carlo simulations at finite temperature could be suitable tools to test our semi-quantitative prediction.	2014-10-01T00:00:00.000Z	2006-07-06T00:00:00.000	{ "year": 2006, "sha1": "15a039ecd4a79b34a07effb7432bc944af78377b", "oa_license": null, "oa_url": "http://arxiv.org/pdf/cond-mat/0607160", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "c3124b63c224c2379566fc419c07a7d989c4ff9c", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
119213729	pes2o/s2orc	v3-fos-license	Ferromagnetic Mott State in Twisted Graphene Bilayers at the Magic Angle We address the effective tight-binding Hamiltonian that describes the insulating Mott state of twisted graphene bilayers at a magic angle. In that configuration, twisted bilayers form a honeycomb superlattice of localized states, characterized by the appearance of flat bands with four-fold degeneracy. After calculating the maximally localized superlattice Wannier wavefunctions, we derive the effective spin model that describes the Mott state. We suggest that the system is an exotic ferromagnetic Mott insulator, with well defined experimental signatures. Introduction.− Mott insulators describe materials that exhibit insulating behavior as a result of strong local interactions [1]. In those systems, strong on site repulsion penalizes the kinetic energy for electrons to hop between sites, rendering the electronic orbitals localized. The strong degree of localization of the electronic wavefunction favors antiferromagnetic alignment of the spins due to Pauli principle [2]. Recent experiments [3,4] indicate that twisted graphene bilayers have a Mott state with an activation gap of ∆ ≈ 0.3 meV that undergoes a metal-insulator transition in the vicinity of a superconducting phase [4,5]. This system is purely made of carbon atoms, with additional degrees of freedom inherited from graphene [6]. That has motivated the question of whether the observed state could be described by a novel Mott insulator [7] or other exotic correlated states [8][9][10][11][12]. Unveiling the nature of the insulating state may be key to explain some of the the remarkable properties in the metallic phase. By twisting two graphene sheets at a small angle of the order of θ ∼ 1.1 • , what was dubbed a "magic" angle, interference due to hopping between the layers leads to a Moire pattern and to a significant reconstruction of the mini bands in the Moire Brillouin zone, which become flat [13,14]. Those flat bands have four-fold degeneracy, which is reminiscent of the valley and spin quantum numbers of the graphene sheets. In general, the confinement of interacting Dirac fermions in flat bands is expected to create an emergent SU(4) symmetry, as previously predicted in graphene heterostructures [15][16][17] and in graphene Landau levels [18][19][20][21][22][23][24]. Here, the Moire pattern forms a superlattice of quasi-localized states with the size of the unit cell set by the twist angle, as shown in Fig. 1. In this Letter, we show that the low energy Hamiltonian of the flat bands at quarter filling maps into the ferromagnetic spin exchange Hamiltonian on a honeycomb superlattice, where S i is the localized spin on a superlattice site i, τ i = (τ x , τ y , τ z ) ≡ (τ ⊥ , τ z ) is an orbital pseudospin operator that is reminiscent of the valley quantum numbers, and J ij > 0 is the exchange coupling. The parameter η ij = −1 when i, j belong to the same sublattice, in which case the exchange interaction has SU(4) symmetry, and η ij = 1 otherwise, including nearest neighbor (NN) sites. This Hamiltonian acts in the Hilbert space which is spanned by four degenerate states per site, \|α, σ , with α = ± and σ =↑, ↓ for the two orbital pseudospins and spin quantum numbers respectively. The existence of direct exchange ferromagnetism in an insulating state is uncommon [15] and reflects the very unusual shape of the Wannier orbitals in this system. Ferromagnetism has been recently observed in insulating van der Waals heterostructures of magnetic chromium trihalide materials, CrX 3 (X=I, Br, Cl) [25][26][27], which have crystalline field anisotropies that produce an ordered Ising state. To the best of our knowledge, we are not aware of any examples of ferromagnetic Mott states which do not involve orbital ordering via a superexchange mechanism [28,29]. After performing calculations of the maximally localized Wannier orbitals of the Moire superstructure, we establish the parameters of a minimal interacting tight- binding model that captures the Mott physics near the magic angle. We show that even though the orbitals are well localized in the Mott regime at quarter filling, surprisingly the direct exchange interaction between different sites is dominant and favors ferromagnetic spin order at zero temperature. While charging effects [30,31], which were not taken into account, may change our conclusions, the scenario of zero temperature ferromagnetism in twisted graphene bilayers seems in line with the reduced degeneracy of the Landau levels measured with Shubnikov de Haas experiments near quarter filling [3]. We discuss the experimental signatures of this state. Bloch Hamiltonian.− The free Hamiltonian for twisted graphene bilayers can be constructed at the lattice level using a parametrization for the hopping amplitudes between sites in the two different sheets, where H g is the graphene Hamiltonian and H ⊥ is the interlayer hopping between the two sheets in real space. The Moire pattern can be used to construct Bloch states that are periodic in the superlattice vectors T i . For commensurate structures, the Moire lattice vectors are parametrized by two integers m and r, and correspond to the twist angle cos θ = 1 − r 2 /2(3m 2 + 3mr + r 2 ), or equivalently θ ≈ r/ √ 3m for small angles. In a basis for Bloch states Φ k,σ ≡ (\|ϕ (1) k,A,σ , \|ϕ (1) k,B,σ , \|ϕ (2) k,A,σ , \|ϕ (2) k,B,σ ) (3) defined in the two sublattices A and B of each of the two layers (1, 2), the Bloch Hamiltonian of the twisted system satisfies H k (r, r + T i ) = H k (r, r )e −ik·Ti . In that basis, are the matrix elements of (2), with a, b indexes running over the four components of basis (3). The hopping amplitudes t ab (r, r ) = cos 2 θ z V σ (r − r ) + sin 2 θV π (r − r ), where cos θ z = d/ d 2 + (r − r ) 2 with d the distance between the planes. V σ (r) and V π (r) are Slater-Koster functions [32], which decay exponentially and were parameterized following previous ab initio works [33,34]. Diagonalization of the Bloch Hamiltonian results in a set of four-component Bloch eigenspinorsψ n,k (r) ≡ r\|ψ n,k that satisfyψ n,k (r + T) =ψ n,k (r)e ik·T and correspond to the energy spectrum ε n (k). We calculate the bands for a small twist angle of θ = 1.0845 • (m = 30, r = 1) near the experimental magic angle θ 0 ∼ 1. At that angle, the Bloch Hamiltonian is a N s × N s matrix with N s = 11164 sites inside the Moire unit cell. The low energy bands (n = 1, . . . , 4), shown in Fig. 2b, are four-fold degenerate at the K points (excluding the spin). They have a two-fold degeneracy at the other two high symmetry points of the Brillouin zone, Γ and M , where they open up a gap between particle and hole branches. At the Γ point, the Bloch states have C 3 and C 2y symmetry, which involves at π rotation around the y axis placed half-way between the two layers (shown in Fig. 1b). We also find numerically that all Bloch eigenspinors satisfy the time reversal symmetry (TRS) relation Tψ n,k (r) =ψ * n,−k (r), with k measured from the center of the Moire Brillouin zone at Γ . The K and K points are hence related by TRS, and must have opposite π Berry phases. This fact indicates that the Bloch states of the twisted structure do not suffer from Wannier obstructions [35], and hence could be reconstructed through a proper basis of Wannier states. Wannier orbitals.− From the Bloch states of the four low energy bands, one can extract the Wannier wave functions in the Moire unit cell, where R is the center of the Wannier orbitals and U nν (k) some 4 × 4 unitary transformation. The four component Wannier spinorsŴ ν (r − R) ≡ r\|Rν are not unique since adding a phase to the Bloch state e −ik·rψ nk (r) corresponds to a new set of Wannier orbitals. We choose the set of maximally localized Wannier orbitals in finding the unitary transformation that minimizes their spread, The minimization was carried with the Wannier90 package [36]. The momentum space k mesh points are generated by the reciprocal supercell lattice vectors with 300 × 300 grid points using periodic boundary conditions, including all high symmetry points. Following the symmetry arguments outlined in Ref. [37], we perform the minimization of the spread enforc- ing the C 3 and C 2y symmetry for the Bloch states around the Γ points. Those two symmetries describe a D 3 point symmetry group, which is a local symmetry of the lattice at AA site regions when the two graphene layers are rotated around a site [38], as depicted in Fig. 1b. In agreement with earlier results [37,39], the Wannier functions that satisfy those symmetries have three sharp peaks centered around either the AB or BA sites, forming a honeycomb superlattice with two-fold degenerate orbitals per site, as shown in Fig. 3. On a given Moire unit cell, we label the Wannier orbitals by the four-component spinorsŴ ν = (w ν,1 , w ν,2 , w ν,3 , w ν,4 ) T . Among the four orbitals,Ŵ ν (r− R j ), two are centered at R j ∈ AB sites and are eigenstates of the C 3 rotation operator, with eigenvalues = e 2πi/3 and * . The other two are centered at R j ∈ BA sites and also have the same eigenvalues and * . From now on, we will label the Wannier orbital spinors based on their C 3 rotation eigenvalues, C 3Ŵα (r − R j ) = e α2πi/3Ŵ α (r − R j ), with α = ± and R j ∈ AB or BA. The two degenerate orbitals centered at a given superlattice site R j are related by TRS, . Orbitals in NN superlattice sites R i and R j are related by the C 2 rotation, Tight binding Hamiltonian.− The effective lattice model of this problem can be constructed by rewriting the Bloch Hamiltonian (4) into a kinetic energy term of the form where R i indexes the sites of the honeycomb superlattice, R ij ≡ R i − R j and the d α (R) annihilates an electron with orbital of type α and spin σ at a given superlattice site. The hopping matrix elements between superlattice sites can be extracted from the matrix elements of Hamiltonian (2) in a basis of maximally localized Wannier functions, Due to the translational invariance of the superlattice, t αβ (R) = 0, α\|H\|R, β . For NN sites, we find that \|t αα \|(1) ≈ 0.384 meV whereas for n-th NN sites t α,−α (n) = 0. Hence, hopping between sites conserves the orbital pseudospin quantum number α = ±. \|t αα \|(n) has a non-trivial dependence with the distance between sites (see table I), in qualitative agreement with the findings of Ref. [37] for a significantly larger twist angle. The Coulomb interactions between lattice sites can be written as is the electron density and κ ≈ 5 the dielectric constant of twisted bilayers encapsulated in boron nitride. We can rewrite this term in terms of d α,σ operators by expressing the density ρ(r) = σΨ † σ (r)Ψ α (r) in terms of field operatorsΨ σ (r) = α,jŴ α (r − R j )d α,σ . The resulting Coulomb Hamiltonian has a direct term and also an exchange part, H C = H d + H e . The first term, with n α (R) = σ d † α,σ (R)d α,σ (R) the density operator and repeated α, β indexes to be summed. The Coulomb coupling is cast as an overlap integral of Wannier orbital is the exchange coupling between lattice sites. In general, we find that the combinations J αβ,βα (R ij ) = Table I: Electronic hopping amplitude \|tαα\|, direct Coulomb interaction V αβ and exchange interaction J αβ for various nearest neighbor sites: on-site (0), and n-th nearest neighbors (n), with n = 1 − 6. Energies in meV calculated for κ = 5. n = 1, 3 and 4 correspond to sites in opposite sublattices. Jαα ≈ ±Jα,−α, with +(−) for sites in the same (opposite) sublattice. J αβ,αβ (R ij ) = 0 for α = β, within the numerical precision. That includes the on site exchange (Hund's coupling), which is zero due to the orthogonality between same site Wannier spinors [15,39]. From now on, we define the only non-zero combination J αα,ββ ≡ J αβ . The numerical values of the hopping energy, Coulomb interaction and the exchange interaction for n-th NNs, is shown in table I, which is the first main result of the paper. We find the on-site Hubbard U αβ ≡ V αβ (0) = 21.2 meV, which is much larger than the first NN hopping t(1), and hence the ratio U/t(1) ∼ 55 falls comfortably in the realm of the Mott regime. The exchange interaction for first NN sites (n = 1) is \|J αβ (1)\| ≈ 5 meV. In general, the diagonal terms J αα (n) > 0 are positive definite, whereas the off diagonal ones can be either positive or negative, J α,−α (n) ≈ ±J αα (n), with + (−) for i, j sites in the same (opposite) sublattice, as shown in table I. For sites in the same sublattice, the fact that J αβ (n) ≈ J αα (n) > 0 is the same for all four combinations of α, β = ± indexes hints at an emergent SU(4) symmetry between spin and orbital degrees of freedom at quarter filling. For sites in opposite sublattices, the exchange interaction has SU(2) symmetry in the spin. It has also both ferro (J αα > 0) and antiferromagnetic (J α,−α < 0) correlations in the orbital sector, depending on the orientation of the pseudospins. Since Hund's coupling is zero, at quarter filling the lower flat bands are in the unitary limit [40], with each Moire superlattice site R j being singly occupied and having a well defined spin σ and orbital quantum number α = ±. Mapping the exchange term in terms of operators, the result is the ferromagnetic exchange interaction announced in Eq. (1), with J ij ≡ J αα (n) > 0 [41]. This Hamiltonian favors ferromagnetic alignment of the spins at zero temperature (T = 0). In the orbital sector different states are possible, including canted magnetism with ferromagnetic order in the pseudospin τ z component, accompanied by staggered (antiferromagnetic) order in the transverse, τ ⊥ direction. The superexchange interaction follows from second order perturbation theory in the hopping energy [42,43] and has the same form as the exchange term in Eq. (1) for η ij = −1 [15]. The superexchange term has SU(4) symmetry and favors antiferromagnetic alignment between nearest neighbor sites due to Pauli principle. It's coupling J → −t 2 /U ≈ −0.01 meV is very small compared to the exchange one, and can be safely igonored. Ferromagnetic Mott state.− Mott-Hubbard insulators have strongly localized states and are known to be overwhelmingly antiferromagnetic due to strong superexchange interactions (t 2 /U J) [44][45][46]. Ferromagnetism occurs mostly either in metallic systems or in metallic bands hybridized with localized moments via the Anderson impurity mechanism [44,45,47]. Within the Hubbard model framework, the only credible mechanism for spin ferromagnetism exists for multi-orbital systems in the context of the Kugel-Khomskii model [42,46], where superexchange can become effectively ferromagnetic in the presence of staggered orbital ordering. We conjecture that the flat bands in twisted graphene bilayers are in a way intermediate between ferromagnetic bad metals and antiferromagnetic Mott-Hubbard insulators. Due to the exotic shape of the Wannier orbitals, the hierarchy between hopping, direct exchange and the local Hubbard interaction, t J U, leads to an anomalously small superexchange. In the charge sector the Mott gap is also anomalously small, ∆ ∼ 0.3 meV W U , where W ∼ 5 meV is the bandwidth, and the system undergoes an insulator-metal transition at T ≈ 4K [3]. In spite of the fact that U/t is large, the strong overlap between the orbitals found in the non-interacting theory suggests that the system is potentially close to an insulator-metal transition [1] due to a charge fluctuation mechanism which presently is not well-understood [30,31]. Nevertheless, the effective spin model we propose in this work should not depend on the details of this mechanism, as long as the system remains quarter filled and does not undergo a charge-ordering transition (potentially accompanied by dimerization) due to Coulomb interactions. In carbon lattices, which are notoriously stiff [48], charge density wave instabilities are hindered by the high elastic energy cost for the system to deform the lattice and restore charge neutrality. Experimental signatures.− Since the honeycomb superlattice is not frustrated, it will exhibit ferromagnetic spin order at T = 0 in the universality class of the ferromagnetic (spin S) Heisenberg model. It is well known that the magnetization M , correlation length ξ and the spin susceptibility χ exhibit peculiar features in two dimensions, since for any T = 0 the system is disordered, with zero Curie temperature. The model has been extensively studied both in zero and finite external magnetic field H on various lattices [49][50][51][52]. At finite field H = 0, M (H) is finite and strongly temperature dependent. In the regime T /J 1, which can take place for T ≈ 2K (where T /J ≈ 1/25), a weak magnetic field of H ≈ 0.2T (i.e. H/J ≈ 1/250) already provides nearly maximum magnetization [50,51]. The susceptibility χ(H) is zero for T = 0 and H = 0 and exhibits a characteristic finitetemperature peak at T = T χ which scales in a welldefined way with external field. It has been established experimentally that doping away from the Mott insulating phase leads to metallic (and even superconducting) behavior [3,4]. Therefore the structure of the ground state and excitation spectrum of this unconventional metallic state is of great experimental and theoretical interest. A profound new feature has emerged at finite magnetic field, which persists both in weak (Shubnikov-de Haas oscillations) and strong field limits (Quantum Hall effect), for hole doping [3,4]. Those measurements suggest a small Fermi surface that develops from doping the correlated insulating phase, accompanied by a possible symmetry breaking of yet unknown origin. The resulting state has a fermionic degeneracy of 2, indicating a reduction of the original four-fold band degeneracy by a factor of 2. This behavior is consistent with the system being in the proximity to a ferromagnetic Mott state, in which the spins align when nudged by an infinitesimally weak field. At the same time, any long-range order in the orbital sector is expected to be much more fragile and disappear quickly due to charge disorder and motion of holes in the metallic state. Therefore we conjecture that in the weak field limit, the ground state emerging from doping the ferromagnetic insulator would be a ferromagnetic, spin-polarized, strongly-correlated metal, with the orbital pseudospin symmetry preserved. Conclusions.− We have derived the effective spin model that describes the Mott phase of twisted graphene bilayers at the magic angle. After calculating the maximally localized Wannier wavefunctions from the lattice, we propose that the system forms a novel ferromagnetic Mott state at quarter filling, with clear experimental signatures. Note added.− After the submission of this work, we became aware of Ref. [53], which also found a ferromagnetic ground state using different arguments. We thank O. Vafek for pointing it out. The linear size of the superlattice cell is The Moiré period L is related to the twist angle, In our numerical calculations, we use m 0 = 30 and r = 1 which correspond to θ = 1.0849 • , L = 129.9196Å, and N = 11164. The superlattice vectors and the corresponding reciprocal superlattice vectors are The two nonequivalent Dirac points are The existence of a commensuration depends only on the relative rotation of the lattice vectors of each layer, and not on the structure of the unit cells of each layer. These amount to different choices of initial basis vectors within each cell. [2] MICROSCOPIC HAMILTONIAN OF GRAPHENE BILAYER It is convenient to enumerate the sites in the sublattice in each layer using two integer valued vectors j = (i, j) and x = (n, m), where j labels the positions of the supercell and x enumerates the sites in the supercell. Then, the Hamiltonian has the form where T j = iT 1 + jT 2 , and c † σ (i, x al ) is the creation operator of an electron with the spin σ on the sublattice a l (= A l , B l ) in the supercell i in the position x on layer l. Note that the Hamiltonian is invariant with respect to translation by T 1,2 , so we can perform the Fourier transform where N s is the number of the supercells. Note that so that we have the Bloch Hamiltonian Introducing the N -component operator k,B,σ , ϕ k,A,σ , ϕ we have the Hamiltonian in the form H = 1/N s kσ x,y Φ † kσ (x) H k (x, y) Φ kσ (y), with the matrix elements indicated in Eq. (5) of the main text. The energy spectrum ε nk and the corresponding Bloch state \|ψ n,k,σ 〉 with n band indices can be obtained by diagonalizing the above Hamiltonian matrix: y H k (x, y) 〈y\|ψ n,k,σ 〉 = ε nk 〈x\|ψ n,k,σ 〉. Note that the Bloch wavefunctionsψ n,k,σ (r x ) = 〈x\|ψ n,k,σ 〉 satisfŷ and the orthogonality Since the periodicity of the Bloch wavefunctions, we will use r ≡ r x + T j for the atomic positions in the entire lattice. In our calculation, the momentum mesh points (k 1 , k 2 ) is generated by the reciprocal superlattice vectors G 1 and G 2 by k = k 1 G 1 + k 2 G 2 using periodic boundary condition, and we used 300 × 300 grid points including Γ, K 1 , and K 2 in the mini Brillouin zone. WANNIER FUNCTIONS The Wannier function centered at R associated with an energy band n is However, the Wannier functions are not unique, so given one set of of Bloch orbitals, another equally good set is obtained from where \|u nk 〉 = e −ik·r \|ψ nk 〉 is the periodic part of the Bloch function, and U nµ (k) is a unitary matrix that mixes the bands at k: The transformation does not preserve the individual Wannier centers, but does preserve the the sum of the the Wannier centers, modulo a lattice vector T. Therefore, the general Wannier function can be written as or, we can define W µ (r − R) as a localized function at R: Note that the Wannier function translated by T i is orthogonal: Sincê we have As a measure of the delocalization or spread of the Wannier functions, Ω[U (k)] = µ 〈r 2 〉 µ − 〈r µ 〉 2 with 〈r µ 〉 = 〈0µ\|r\|0µ〉 and 〈r 2 〉 µ = 〈0µ\|r 2 \|0µ〉, the maximally localized Wannier functions can be obtained by minimizing Ω with respect to U nµ (k) [8]. SITE-SYMMETRY GROUP AND SYMMETRY-ADAPTED WANNIER FUNCTIONS Specifying a set of sites {R 1 , R 2 , · · · }, where the Wannier functions will be centered, the site-symmetry group of a given R 1 , denoted by G site , is a subgroup of the full symmetry group of the lattice, G whose elements leave R 1 unchanged, so g site = (R s \|T s ) ∈ G site satisfy where R s , T s are the rotation and the translation part of the symmetry operation. The full symmetry group G can be decomposed into left cosets of the subgroup G site as G = j,n g jn G site , g jn = (R j \|T j + T n ). (28) Here, g j0 (T 0 = 0) is one of the symmetry operations that maps R 1 to its symmetry-equivalent point R j as Since j = 1 corresponds to the original point R 1 , g 10 = (E\|0), where E denotes the identity operation. From the site-symmetry group for a given R 1 , the site-symmetry adapted Wannier functions centered at R 1 are defined as the basis functions of the irreducible representations of the site-symmetry group G site . These Wannier functions are represented asŴ where α = 1, 2, · · · , n α denote the component of the basis functions with n α the dimension of the irreducible representation. The Wannier functions transform aŝ FromŴ α1 (r), we can generate Wannier functions centered at R j aŝ In our model,ĝ site =Ĉ 3 and R 1 is a site BA, leading toŴ 1 ( For R 2 a site AB, we haveŴ 1 (r 2 ) =Ŵ − (r 2 ) =Ŵ − (r − R 2 ) andŴ 2 (r 2 ) =Ŵ + (r 2 ) =Ŵ + (r − R 2 ) with R 2 = C ′ 2y . From these symmetry-adapted Wannier functions W αj (r), one can construct the Bloch functions Ψ k,αj (r) aŝ These site-symmetry adapted wannier functions and the Bloch functions transform aŝ where The index j ′ is determined by g and g j to maintain T j ′ j being lattice vectors. Then the irreducible representation of G is given by The unitary matrix (22) for the maximally localized Wannier functions with the site-symmetry adapted can be obtained by solving whered n,n ′ (g, k) = drψ † n,Rk (r)ψ n ′ ,k (g −1 r) (40) withψ nk (r) the eigenfunctions of the Bloch Hamiltonian. Here, we used the notation, µ = {αj}. With these D α ′ j ′ ,αj (g, k), d α ′ α (R s ), andd n ′ ,n (g, k) as inputs for the Wannier90 package [10], we can find the site-symmetry adapted Bloch wavefunction, Eq. (34), asΨ leading to the site-symmetry enforced Wannier functions, Eq. (35). Hopping amplitude The most general tight-binding Hamiltonian is of the form where the creation operator d α (R j ) is associated with the Wannier functions Then the hopping amplitude t αβ (R jj ′ ) is given by where T jj ′ = T j ′ − T j , with T j the lattice vector of the cell enclosing R j . Interaction Hamiltonian The exchange interaction follows from the Coulomb interaction where ρ(r) = σ Ψ † σ (r) · Ψ σ (r) are density operators written in terms of field operators where is the exchange coupling. We find numerically that only the J ij,ααββ ≡ J ij,αβ combination is non-zero. The exchange interaction has the form Now mapping the d operators into spin and pseudo-spin operators [9,11], with α = ± and σ = ± and τ α = τ x + iατ y , S σ = S x + iσS y , (47) can be cast into the form NUMERICAL DETAILS Our calculations are carried out with the tight-binding Bloch Hamiltonian of the microscopic twisted graphene bilayer. The maximally localized Wannier orbitals were found with the Wannier90 package. The momentum space k mesh points were generated by the reciprocal supercell lattice vectors G 1 and G 2 by k = k 1 G 1 + k 2 G 2 with 300 × 300 grid points using periodic boundary conditions. This way, the Brillouin zone includes Γ, K 1 , and K 2 high symmetric points. We modified the code in the package for the Brillouin zone to include all the k points and the corresponding −k points (modulo a reciprocal lattice vectors), reflecting the time-reversal conjugates of the Bloch wavefunctions. The first two input functions are eigenstates of the C 3 operator with eigenvalues , while the other two with eigenvalues * .	2018-12-10T17:30:53.000Z	2018-12-06T00:00:00.000	{ "year": 2018, "sha1": "ec0ffd9da6f11c3d699e267042f2345a2b68717d", "oa_license": null, "oa_url": "http://arxiv.org/pdf/1812.02550", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "ec0ffd9da6f11c3d699e267042f2345a2b68717d", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics", "Medicine" ] }
238681439	pes2o/s2orc	v3-fos-license	Tips for a reduction of false positives in manual RT-PCR diagnostics of SARS-CoV-2 : RT-PCR is the standard gold technique for testing the presence of RNA of the coronavirus causing Severe Acute Respiratory Syndrome (SARS-CoV-2) due to its high specificity and sensitivity. Despite its general use and reliability, no lab in the world is immune to the generation of false positives. These errors cause a loss of confidence in the technique's power and damage the image of laboratories. More importantly, they can take a toll on tested individuals and have economic, psychological, and health-associated effects. Most false positives are caused during a manual operation inside the laboratory. However, not much has been published about the errors associated with particular laboratory techniques used to detect the virus since the beginning of the actual pandemic. This work precisely reflects on events that occur during manual RT-PCR diagnostics in a COVID-19 laboratory, providing tips for reducing false-positive results. Introduction In COVID-19 diagnostics labs, most errors in reporting consist of false negatives due to the low viral load that escapes detection. Therefore, the sampling time is significant: a sample taken too early after the person has been infected may have too low a viral load to be detected 1 . Likewise, a sample taken after the patient has cleared the virus almost wholly would result in a very low viral titer. The skills of the doctor or nurse in charge of taking the samples from patients can also contribute to generating false negatives. In addition, the choice of transport medium where the sample is carried to the labs for analysis, whether the sample contains blood, and the temperature during storage and/or transportation can make a difference in the quality of the sample before its processing. False negatives are of great concern for public health management, but the impact of false positives has come to the fore recently due to the distress they can cause in the lives of patients and public health. The RT-PCR (Reverse Transcriptase Polymerase Chain Reaction) is the standard gold technique for detecting the SARS-CoV-2 virus. It is routinely used for samples from diverse origins, such as nasopharyngeal and throat swabs, sputum, broncho-alveolar lavages 2 , and anal swabs 3 . It can detect as few as 5 copies of the virus in a sample 4 and is, therefore, the technique of choice to reveal which individuals are contagious at any time of their infection period. They outcompete serological tests, which need a higher viral load to produce a positive result or indirectly measure the virus's presence in the organism. The RT-PCR has a sensitivity (ability to detect true positives) of 70% and a specificity (ability to report well true negatives) close to 100% 5,6 . Having such sensitivity, the technique can easily amplify contaminating virus particles that don't belong to the actual samples to be tested. Hence, RT-PCR can generate false positives with a significant likelihood. Although few studies report on rates of false positives for COVID-19, some estimates of false-positive rates suggest them to be in a range of about 0.3-4% [7][8][9] . For SARS-CoV and MERS-CoV, the rate oscillated between 0.3-6.9% 10 . False positives can also arise due to errors in reporting the results and uncertainties regarding the cycle-threshold (Ct) value used as a diagnostic criterium 11 . The former point can be addressed by judicious data entry; however, the latter poses a true challenge when Ct values are close to the cut-off. Here, the likelihood of a false positive or false negative is highest. For instance, if a Ct of 40 is chosen as the diagnostic criterium, does a Ct of 39.5 necessarily mean that 1) the true Ct is indeed below 40 and 2) is the patient still infectious? 11,12 A false positive report on healthy people can have dramatic consequences, varied and challenging to quantify. A politician may be deprived of a critical meeting for the citizens he represents. A skilled celebrity may be deprived of a significant sports competition. Worse even, the economic distress caused to those belonging to low-income groups in the society is more significant, for they may have to stop working, may have dependents, and no savings. The stigma of being called positive and the fear of suffering complications can also be detrimental psychologically for many. Besides, healthy but wrongly diagnosed positive, people may be put at risk of real contagion when moved to areas in a hospital with infected patients or may suffer the delay of an essential medical procedure 10,13 . Unfortunately, due to the severity of the current pandemic, it is not feasible to perform confirmatory PCRs for every patient whose sample yielded a positive result. Instead, current guidelines by international and national public health agencies recommend evaluating every PCR result on a case by case basis in combination with the evaluation of local infection rates, clinical signs and symptoms, lung CT scans, and history of exposure 10 . The consequences for the labs that report false positives and negatives can also be dramatic. With their image damaged, contracts for private labs may be postponed, and potential customers' confidence can vanish. Public labs are also questioned when reports of false positives appear. However, false positives are unavoidable and cannot be eliminated in any laboratory. However, it should be possible to reduce them to a minimum to better comply with the targets of national regulatory agencies. The standard laboratory workflow for COVID-19 testing goes in one direction and has multiple barriers to prevent cross-contamination. Laboratory technicians have routinely tested themselves for covid, for apparent reasons. Still, the laboratory is the primary source of false positives, mainly cross-contamination due to human error 10 . In this review, we'll discuss working habits that reduce the occurrence of false-positive reports of coronavirus infection while navigating the manual processing of nasopharyngeal swabs in all phases of the workflow. The tips provided are based on personal experience of false positives while working in a COVID-19 test lab at Yachay Tech University, using typical kits and equipment. Reduction of false positives in the RNA extraction area The manual processing of samples during RNA extraction constitutes the highest source of false positives in the laboratory, but there is much room for improvement. The processing of samples in the RNA extraction area starts with setting the laminar flow hood under UV light for several minutes while the airflow stabilizes 14 . A small centrifuge should be located inside the hood, and it should be left open during UV irradiation to expose the rotor. The space under the hood should not be crowded to help the airflow exert its function and allow the UV light to reach most surfaces. If more than one batch of samples is going to be processed in the day, it is advisable to have rounds of UV irradiation in between batches. Next, parts and equipment to be used to process the samples are cleaned with 70% ethanol. Only filtered tips are used for the extraction and must be changed for every sampling step of the extraction process. Tube racks should be stable to avoid spills. Reagents should be aliquots of the original kit contents to minimize costs if contamination (reagent contamination is a significant source of false positives but easy to identify because most likely, all samples of the same batch would come out positive and with similar Ct values). Lab technicians must wear protective gear with very fit pairs of gloves, without folds on the surface of their fingertips. Once the site is clean, the lab technician starts preparing a master mix that ensures that all the samples get the same initial buffer solution. Fig. 1 shows the process of RNA extraction for a typical commercial kit, with few modifications of the original protocol. Similar steps apply to many other commercial kits. Double arrows A to C at the bottom indicate the times at which different contamination types can occur. The master mix contains an extraction control that helps verify whether the extraction process has been optimal for every individual sample. As an example, this control can be the RNA of a cellular household gene such as actin. This will later appear as a specific curve in the PCR reaction since the primers for that gene will be included in the primer mix used for the PCR reaction. The signal generated during PCR amplification will appear in a different channel than that of the SARS-CoV-2 target gene of interest. The absence of the extraction control curve in all the samples of the batch in the PCR would indicate that it was probably not added to the master mix. If absent in only one or a few samples, it could indicate the presence of inhibitors of the PCR reaction such as ethanol, and the extraction of those particular samples should be repeated. Once the master mix has been added to all the microcentrifuge tubes, it is time to place the nasopharyngeal swab samples inside the hood, which had been kept in a refrigerator upon arrival. Careless manipulation of swab samples can generate mix-ups leading to false positives. The manipulation of patient samples entails a high risk of contamination (arrow A, bottom of Fig. 1) that can then be passed down throughout the extraction process. Tubes should ideally be opened with a hand that does not hold the micropipette to avoid contaminating it. A brief vortex ensures a good mix for every sample and potentially generates aerosols that can contaminate the working area. That is why it is essential to work under the hood, with the airflow removing those aerosols. Aerosols can also adhere to the shaft of the micropipettes, especially when they are introduced deep into the patient's sample tubes or touch the swabs that usually come within. If that happens, the parts suspected of being contaminated should be wiped with 70% ethanol between pipetting samples. Once the micropipette tip is loaded with the sample, keeping it at an angle instead of vertically prevents dripping of the content for a sufficiently long time before adding it to the microcentrifuge tube. Also, it is common practice in molecular biology or microbiology labs to eject the residual volume of the tip with an extra push of the micropipette plunger. This, however, is an essential source of aerosols. Therefore, it is desirable to avoid ejecting that residual volume to avoid cross-contamination during the extraction of SARS-CoV-2 viral RNA. The negative control of the extraction (NCE) must be the last microcentrifuge tube of every batch of samples to ensure that contamination is appropriately detected. Nuclease-free water is added to the tube instead of the viral sample. Being the last tube of the batch, the NCE will help identify cross-contamination events from aerosols of a positive sample of the same batch. In the PCR, the NCE should only yield the curve corresponding to the extraction control. In the case the NCE produced the curve corresponding to the SARS-CoV-2 gene being tested, the extraction of the whole batch should be repeated. Fig. 2A shows the fluorescence profile of the extraction control in the amplification plot of a positive sample. Finally, many extraction kits require a few minutes to inactivate the virus. That time can be used to clean surfaces in contact with patient samples, tip boxes, micropipettes, and the microcentrifuge tubes' exterior. Once the samples have been inactivated in the first steps of the RNA extraction process, the risk of acquiring floating virus particles is significantly reduced, and many labs continue their work on the bench. However, the risk of cross-contamination among samples is higher on the bench than on the laminar flow hood. Hence, although not deemed necessary for the protection of the lab technician, a laminar flow hood helps avoid the generation of false positives among samples. During the intermediate steps of the extraction process, errors during pipetting and handling of samples and reagents account for additional cross-contamination risks (arrow B, bottom of Fig. 1). It is advisable always to assume that aerosols from positive samples are present in the air to increase the sense of alertness and carefulness during the handling of microcentrifuge tubes, buffer solutions, and micropipettes. When using the micropipette, the tips should not enter too deep into solutions because that facilitates the carryover of the solution and its posterior dripping. Very importantly, buffer solutions should be opened only when needed, keeping tubes closed in the meantime. Most extraction kits require the use of small centrifuges for several steps. The centrifuges are challenging to clean from aerosols generated from positive samples during the manipulation of the tubes. A high risk of contamination occurs in the elution step (arrow C, bottom of Fig. 1) when a column is commonly placed inside an open microcentrifuge tube. Having it open, the lid of the microcentrifuge tube sometimes breaks during the final elution spin. Again, it is essential to use tightly fitting gloves when picking up the tubes and columns from the centrifuge. Centrifuges should be cleaned thoroughly after the processing of every batch of samples. The use of automatized systems for RNA extraction improves the processing speed enormously and reduces the chances of cross-contamination. During manual feeding of patient samples to the robot in the laminar flow hood, errors can still happen at the initial step of the process. If possible, the operator should close all the wells that are not being used. To avoid pipetting errors, it is advised to mark during the loading process the wells of the plate where the sample has been already added. Once the work under the laminar flow hood has ended, it must be cleaned, and UV irradiated to inactivate any virus particles that could have escaped in aerosols during the extraction process 15 . Automatized systems also need to be UV irradiated after use. Reduction of false positives in the pre-PCR area The most common method for producing reliable test results involves the use of an RT-PCR thermocycler. The conversion of the extracted RNA into DNA is carried out by a retro transcriptase, while a DNA polymerase does further amplification of that DNA. Both enzymes can be part of the same mix (one-step RT-PCR) or separate reactions (two-step RT-PCR). The PCR mix must include primers and fluorescent probes for a SARS-CoV-2 gene and for the extraction RNA target gene that was added during viral RNA extraction as a control. These reagents are added in the pre-PCR area, space physically separated from the room in which the thermocycler is used. In the pre-PCR area, the lab technician will aliquot the PCR mix in the wells of PCR plates or tubes and then add the extracted RNA and controls. There is absolutely no risk for the lab technicians in the pre-PCR area to be contaminated with active viruses. They can, however, bring contamination to the sample preparation area on her clothes. To avoid carrying over amplicons or viral RNA on their clothes, personnel who work in the pre-PCR area must use a different lab coat and gloves than those used in other areas of the laboratory. Although in principle, the assembling of the PCR can take place on the bench, it is advisable to use a laminar flow hood for that purpose, similar to what was recommended for the extraction area, to prevent aerosols from positive RNA samples contaminating the wells of the PCR strips or plates. In Ecuador, a laminar flow hood is a mandatory requirement for this area. We must never forget how sensitive the PCR technique is and detect as few as 5 copies of viral cDNA 4 . Similar to what is done in the extraction area, the laminar flow hood must be irradiated for 10-20 minutes with UV light until the airflow stabilizes before assembling the PCR reaction. Despite working under the laminar flow hood, we must assume that aerosols containing RNA from positive samples could still be present and cause false positives. Hence, a good piece of advice for the assembly of the PCR is to keep all the wells in the PCR strips or plates covered at all times and only lift the caps (strips) or optical film (plates) when a new sample must be added to the appropriate well. Keeping nearby a log of the position of the samples in the wells helps to prevent mistakes during sample loading. Again, to prevent the spread of aerosols, it is essential to use very fit gloves, without folds on the surfaces of the fingertips, to properly open and cap the tubes (strips) or hand the optical film (plates). The closing of the caps of PCR tubes may require applying great force, which can cause vibrations on the cold rack that supports the strips and make them move or fall off the rack. For that reason, it is good to write a small number or another type of code on top of each PCR strip that indicates the orientation and position of wells in the PCR to be run. Another suggestion to avoid aerosols is to not pipet the sample up and down to mix in with the PCR reagents in each tube. This common practice is not needed since the PCR strips or plates will be spun down in a centrifuge before taking them to the thermocycler. Combined with the high temperatures of the PCR process itself, the homogeneity of the reagent solution is assured. The PCR reaction's assembly, a negative and one positive control, respectively, needs to be included. The negative control (Non-Template Control, NTC) contains the mix with the buffer, polymerase, and primers for the PCR reaction but nuclease-free water is added instead of extracted RNA. The control also lacks the extraction control added to the samples processed in the RNA extraction area. At the end of the PCR run, the NTC profile should be a flat line for all the channels. The positive control (PC) has the same content as the NTC, but instead of water, the same gene of the SARS-CoV-2 virus that the primers detect in the actual patient samples. The PC also lacks the extraction control added to the samples in the extraction process, and it also emits fluorescence in a different channel. Fig. 2A shows the fluorescence profile from a positive sample for SARS-CoV-2 at the end of the PCR run. Both NTC and PC must be placed at the end of the assembly of the PCR reaction after all the other samples have been loaded. In this way, the NTC will serve the purpose of letting us know whether cross-contamination has taken place during the assembly process. If the NTC yields a positive signal for the gene of interest, the PCR should be repeated because either the master mix was contaminated or aerosols from a positive sample were spread during the assembly process to other samples. If possible, the negative controls from the RNA extraction step (NCEs) should not be added at the end to the PCR strips or plates to minimize their possible contamination with aerosols from positive samples during the assembly process. In this way, if the NCEs yield a positive signal for the viral gene of interest but the NTC produces a negative signal, we could be confident that the cross-contamination event has taken place during the RNA extraction steps and not in the pre-PCR area. If, on the other hand, one suspects contamination events taking place in the pre-PCR area, it is advisable to aliquot known negative samples or NTCs to the PCR plate and check afterward whether they yield a positive signal. Although we focus mainly on the events that can generate false-positive results, there is also a chance of generating false negatives in the pre-PCR area. One way is to accidentally not pipetting up any liquid from the RNA sample. This can happen, for example, if there is an air bubble in the microcentrifuge tube and the tip of the micropipette just absorbs air instead of sample. That is why it is important not to lower the concentration during work on the bench and always look at the tips to make sure the liquid is absorbed and again check after centrifugation that every tube in the strip or plate has the same volume. Staying focused and being mindful of every movement of the hands above the wells in the strips or plate helps minimize the risk of contamination at every step of the process. Where available, the use of multichannel pipettes also greatly helps to speed up the PCR assembly process. However, they can also fail to acquire the desired volume in some of the channels and be a source of false negatives. Reduction of false positives in the PCR area An essential source of false positives that can originate in the PCR area is amplicon contamination. This occurs when the products of a previous PCR are accidentally released and migrate to other parts of the laboratory. The tubes with positive samples contain trillions of copies of the amplified genes that could easily contaminate future extraction processes and PCR assemblies. Even an initially slightly positive sample with a low viral load whose manipulation would not represent much risk during the extraction process can cause significant problems once the target genes have been exponentially multiplied 16 . To avoid this type of risk, PCR tubes or plates should be disposed of outside the PCR room at the end of the PCR run. Both the pre-PCR and PCR rooms must be physically separated and have independent air extraction systems. It is recommended not to cross from the PCR room to the extraction and pre-PCR rooms wearing the same protective clothing. Another possible source of false positives occurs when tube strips or plates are not adequately balanced during the centrifugation before the PCR run. This causes vibrations that can disperse the contents of the PCR tubes onto their walls, leading to underestimated readings by the thermocycler. The best way to prevent amplicon contamination is the implementation of a regular cleaning protocol for all surfaces in the lab-based on 10% (w/v) sodium hypochlorite -a chemical amplicon oxidizer-followed by a rinse with 70% ethanol (to avoid the corrosive effect of the bleach on equipment) 17,18 . Commercial products abound that include diluted NaOH in their formulations. The use of UV irradiation in Class II biosafety cabinets 14 or portable UV lights can also be of great help. Inactivating enzymatic methods such as degradation by uracyl-N-glycosylase 19,20 can be included in the PCR reaction mix. On the other hand, inactivation protocols that require the opening of the PCR tubes after amplification are to be avoided. The Ct (cycle threshold) is the cycle number when the fluorescence of the amplicons surpasses the threshold fluorescence during the PCR run. In our lab, based on the experience of other labs using the same diagnostics kits, a Ct value of 40 or earlier is interpreted as a positive result for COVID-19. The Ct value also measures how many copies of the target gene were in the original sample. The smaller the Ct, the more viral copies the sample has. Fig. 2B shows the amplification plots of four samples, two of them with low Ct values indicating high titers of the target gene. A sample with Ct 10 contains thousands of more copies in origin than a sample of Ct 20. Extraction of samples with high viral titers constitutes a vital source of contamination for the subsequent samples in the batch due to the unintended release of aerosols 20 . One can never know beforehand which sample has a high viral titer; this is learned after the PCR. Automatic extraction reduces the risk of cross-contamination from high viral titer samples. In the analysis post-PCR analysis, the lab personnel must decide in every case whether the sample is positive or negative, looking both at the Ct values and the curves. False positives can appear due to the presence of inviable viral particles in patients who are in the process of clearing the remains of the virus from their bodies. If the data look sound, the lab must report those results as positives and let the clinicians make the final decision based on the patients' medical history, local rates of COVID-19 infections, patient's signs and symptoms of the disease, or serological analysis 6,10,21 . When the signal of the SARS-CoV-2 gene in the PCR appears with a Ct value close to (above or below) 40 (like the sample with Ct 39 in Fig. 2B), a definite diagnosis should not yet be made by the lab processing the samples. Such a profile could result from contamination with aerosols of a close sample with a very low Ct value. The fastest solution is to repeat the extraction of that particular sample on the same day. Alternatively, one can request another sample of the same patient within 2-3 days: if the patient happened to be at the beginning of the infection process, she will appear positive in the second PCR, while a negative result would mean clearance of the virus at the end of the infection period. On average, the best time to get a sample that results in a lower chance of false-negative is eight days after infection or three days after the onset of symptoms 1 . All lab members in charge of the analysis post-PCR should be using precisely the same criteria. For that, an algorithm like the one in Fig. 3 can be devised and agreed upon. That way, based on the PCR results, any user can quickly troubleshoot unusual results, find the cause of false positives, and make amendments before reporting the results. Note that despite all the controls being correct at the bottom of Figure 3, one must still be wary of an unusual number of positive samples, especially if they appear grouped. In this case, we recommend that a subset of the samples that also includes negatives be subjected to another round of viral RNA extraction and PCR. As an internal quality control measure, a small fraction of the daily samples arriving at the lab can be randomly selected for retesting. The selection of samples should be made before knowing the results of the PCR test, allowing for the inclusion of both positive and negative samples. Discussion Having navigated through all the steps of manual sample processing in a laboratory of COVID-19 testing, one must not forget the importance of selecting the best personnel for the different tasks to be performed. Good leadership is necessary to be aware of the strengths and weaknesses of every person working in the lab and get the best out of everyone. False positives are mainly produced through unintentional errors during viral RNA extraction, the assembly of the PCR, and even data management. Hence, a tremendous amount of focus and manual dexterity is expected in laboratory workers to avoid pipetting errors. Maintaining focus is essential in the COVID-19 testing laboratory since many everyday tasks can be repetitive, and the operator may easily engage in detrimental mind-wandering behavior for brief periods of time 21 . At times the amount of COVID-19 samples arriving at the lab is too tremendous, and several shifts are required to process them. The schedules must be made to avoid too much strain on the workers, thus ensuring that the workers enjoy what they do and maintain a positive attitude in the lab. Remembering that sample numbers are numbers and represent people and their families can be another source of motivation. It is common practice to interpret PCR results in the context of the pretest probability of the disease 13 . For example, a patient that appears positive for COVID-19 by PCR but has no symptoms or medical history of the disease has no antibodies and was not exposed to the disease could be considered a false positive. In our opinion, that is a call that doctors should make but not be the labs reporting their results. In case of doubt, it is always better to repeat the RNA extraction or PCR or ask for a new sample. Similarly, a positive result in someone already known to have contracted the disease weeks ago is still positive, although she may not be infectious anymore and is probably just shedding inviable virus particles 22 . Samples with low Ct values cause more trouble in the lab in terms of the production of false positives than those of high Ct values during the manual processing of samples. Although sampling introduces a great deal of variability in the first place, low Ct values reflect high titers of the SARS-CoV-2 gene. This seems associated with a high viral load in the original patient samples, for those with low Ct values are more culture-positive than those with high Ct values 23 and correlate with the risk of intubation and in-hospital mortality 12,24 . Hence, it could seem that reporting the Ct value would be very useful for physicians. However, the issue is up for debate nowadays, mainly due to the high Ct values between and within methods 11 and during sampling. Conclusions From our experience, it is virtually impossible to eliminate false positives in the COVID-19 diagnostics lab completely. Manual processing requires multiple manipulations of samples and reagents, which translates easily into events of contamination. Automatized systems can significantly reduce (but not eliminate) the appearance of false-positive results. Contamination from samples with very low Ct values is more likely to occur during the extraction process than during the assembly of the PCR. It is advisable to routinely repeat the extraction of positive samples from a batch in which very high viral titer samples were included, especially when too many positives appear clustered together. Also, as an internal quality control measure to gain confidence in their results, labs can temporarily keep a small fraction of their daily samples, chosen randomly, for retesting. With the comments of this article, we hope to contribute to a reduction in the rate of false positives in labs dedicated to similar tests elsewhere during the manual processing of COVID-19 samples. Declarations Funding: No funds, grants, or other support was received. Conflicts of interest/Competing interests: Francisco Alvarez, Mariela Perez, Marco Gudiño, and Markus Tellkamp declare that they have no conflicts of interest.	2021-09-27T19:02:24.493Z	2021-08-15T00:00:00.000	{ "year": 2021, "sha1": "b883070e832be4ba3adb919a05d8d87f476c377d", "oa_license": "CCBYSA", "oa_url": "https://www.revistabionatura.com/files/2021.06.03.11.pdf", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "73219250c1fe71f060ada8d386a2e4439a0dfd6b", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
226582754	pes2o/s2orc	v3-fos-license	The impact of economic value added (EVA) adoption on stock performance Article history: Received May 1 Introduction It is generally accepted that the normative role of the executive manager is to maximize firm value (Wallace, 1997;Malmi & Ikaheimo, 2003). But a long-standing problem for the owners of the firm has been that a fiduciary role by the manager does not occur naturally. There must be some form of compensation design that induces the required management behaviour. Any design, in turn, requires a measure of firm performance. How do we know that a manager has increased firm value? One traditional measure is Share Price (Jensen & Murphy, 1990). However, Share Price depends on factors beyond management control (Sloan, 1993;Lambert, 1993). Furthermore, in the short run, maximising share price might not always be aligned with maximising the firm's intrinsic value (Mramor & Valentincic, 2001). Other traditional measures used in assessing managers' performance include earnings, return on assets, return on investment and other cash flow measures. However, there are wellknown problems with these measures (subjectivity and ease of manipulation). The search was therefore on for new performance measures that would improve on the traditional accounting-based measures. Practitioners obliged. Consulting firms began marketing their own value based performance measures (Myers, 1997). The most prominent of these was Stern and Stewart's Economic Value Added (EVA) (Stern & Stewart, 1991). Stewart (1994, p.75) proclaimed that "EVA stands well out from the crowd as the single best measure of value creation on a continuous basis", and that "EVA is almost 50% better than its closest accounting-based competitor (i.e. earnings), in explaining changes in shareholder wealth". However, although these measures were driven by practitioners, value-based ideas had been known for more than a century before they were marketed by consulting firms. Wallace (1997) states that Alfred Marshall mentioned the residual income concept in 1890. This same measure was also discussed by Canning (1929) and Preinreich (1937) who referred to it as excess earnings, while Edey (1957) referred to it as super-profits. The basic idea of this measure is that unless a business "returns a profit that is greater than its cost of capital, it does not create wealth; it destroys it." (Drucker, 1995, p.59). Previous research has examined whether the adoption of EVA's incentive compensation plan has any impact on managers' decision making (Wallace, 1997;Kleiman, 1999;Hogan & Lewis, 2005). All of this empirical research has the common assumption that the adoption of the EVA compensation system will rationalize a firm's investment decision and will lead to using the existing assets more efficiently to generate more residual income and, hence, to maximize shareholders' wealth, as well. These studies tested EVA adoption effects for up to five years after adoption. A major limitation in the existing literature is that the focus was on changes in the manager's behaviour as reflected by accounting fundamentals. Adoption of EVA is supposed to increase asset dispositions, repurchases, and dividends, while at the same time decrease new investments and accounts payable. However, if adopting EVA really led to optimal management decisions, the above effects may be necessarily observed. For example, a firm could have under-invested prior to adopting EVA. An optimal behaviour would see an increase in new investments rather than a decrease. Thus, even if EVA did induce the management to optimise firm value, the optimal change in the accounting variables may not always be in the same direction. This is probably why the various studies that studied the effect of EVA adoption have found different and often conflicting results (Wallace, 1997;Kleiman, 1999;Cahan et al., 2002;Hogan & Lewis, 2005;Balachandran, 2006). In short, while existing evidence points to a change in management decision, there is no evidence that adopting EVA increases value. Apart from management-behaviour effect studies mentioned above, the other line of enquiry focused on value relevance (Biddle et al., 1997;Lehn & Makhija, 1997). The general finding is that EVA actually has poorer explanatory power than other traditional performance metrics such as earnings. Surprisingly, we know of no study that has tested directly the value effect of EVA adoption. The aim of this paper is to fill this gap in the literature by showing whether or not EVA adoption leads to a significant increase in firm value as reflected by its market prices in the long run. While we agree that in the short run, the market might be subject to a number of behavioural and informational inefficiencies. There is growing evidence that the stock market does not incorporate all firm information into the stock price quickly and completely. Therefore, the critique that contemporaneous association between price and EVA does not reflect reality is likely to be correct. However, we take a different take. Our basic contention is that although prices adjust slowly to information, long horizons are sufficiently long for markets to incorporate almost all relevant information into prices. In other words, our basic assumption is that markets are efficient in the long run (Fama, 1998, JFE, 49, 283-306). This paper will therefore examine whether the adoption of the EVA framework enhances the firm's performance and to gauge the long-term effects of such an adoption on the firm's value. It also assesses whether the market reacts to the announcement of the adoption of EVA as a compensation system. The event study methodology initially introduced by Fama et al. (1969) will be used to assess the impact of EVA's adoption on a firm's performance. The adoption the trademarked EVA performance measure grew rapidly during the 1990s in the USA, but does not seem to be popular in Europe. For example, in the UK only four companies have reported the EVA as a performance and management tool. 1 The structure of this paper is organized as follows. Section 2 summarises the main results of the previous studies that have investigated the impact of EVA adoption on a firm's performance. Sections 3 and 4 describe the sample and the methodology respectively. Section 5 discusses the empirical results obtained and finally section 6 summarises the main conclusions. Previous studies To our knowledge, there are only two studies on stock price performance following adoption of EVA as a compensation measure. The first study focuses on the short-term market reaction to EVA adoption. Tortella & Brusco (2003) used a sample of 65 EVA adopter firms and compared the daily abnormal return of adopting firms to that of two index portfolios. They used a window of 30 days prior to the adoption and 100 days post the adoption. Their results indicate that the daily cumulative average abnormal return is insignificant throughout the post event window. They conclude that the market does not react to the adoption of EVA in the short run. Ferguson et al. (2005), on the other hand, do a long-term event study on EVA adoption. They use a list of 65 EVA adopter firms provided by Stern Stewart & Co. However, our study differs in many respects. First, we use a larger sample of 89 firms, which is likely to improve the statistical credence of our empirical results. Moreover, our sample goes to 2001 while their sample ends in 1998. Second, they rely exclusively on the cumulative abnormal return (CAR) concept in their analysis. Abnormal returns have their own shortcomings and have been criticised severely for prompting certain biases Lyon, 1996, 1997;Fama, 1998). In this study we use both CAR and the buy-and-hold abnormal return (BHAR) concept. BHAR is well known for its ability to represent investor's experience (Mitchell and Stafford, 2000). Third, they use market model to estimate predicted or expected returns. This may suffer from the bad model problem (Fama, 1998). In this study we use matching control firms to calculate abnormal return or abnormal buy and hold return. The use of matching control firm is less likely to suffer from the bad model problem (Stuart, 2010). The other line of enquiry that is closely related to our objectives is the one that focused on the effect of EVA adoption on internal firm decisions such as financing and operating decisions. This kind of research was initiated by the seminal paper of Wallace (1997), and was then followed by several studies over the following decade (Kleiman, 1999;Cahan et al., 2002;Hogan & Lewis, 2005;Balachandran, 2006). The underlying assumption in the above studies is that firms adopting EVA will create the ability to enhance their profitability and maximize shareholder's wealth. This can be achieved by increasing a firm's ability to generate a large residual income and encourage managers to invest in those projects that can earn more than the cost of the capital invested. Furthermore, EVA's proponents claim that the adoption of the EVA framework will affect the manager's behaviour and lead to the best alignment of management interests with those of the shareholders (Stewart, 1991;Wallace, 1997). The existing empirical studies on the adoption of residual income-based performance incentives have found mixed results. Wallace's (1997) study initiated this line of enquiry by addressing the changes in a number of firm decisions following the adoption of residual income or EVA measures. Wallace (1997) was been replicated by a number of scholars such as Kleiman (1999), Hogan & Lewis (2005) and Balachandran (2006). Wallace (1997) compared a group of forty companies adopting residual income or EVA as a compensation plan with the same number of control firms to examine whether the adoption of these measures, impacted the investing decisions, financing decisions, operating decisions and shareholder wealth. Kleiman (1999) extended Wallace's sample to 71 firms, but focus exclusively on firms adopting EVA as an incentive compensation system. Kleiman (1999) found that EVA adoption led to higher stock return performance. Inconsistent with Wallace (1997), Kleiman's (1999) results do not show any capital expenditure decline following adoption of EVA. On the other hand, he reports that EVA-adopting companies significantly increase their financial leverage, extend share repurchases, and enhance both operating margins and operating profits before depreciation. Hogan and Lewis (2005) used a sample of 108 firms that chose to adopt the economic profit plans (EPPs) as incentive compensation systems between 1983 and 1996 to examine whether the adoption would affect these firms' operating, organizational, financial and compensation characteristics. The findings of Hogan & Lewis (2005) reveal that EPP adopter firms in general show a significant enhancement in operating performance relative to their past performance (pre-adoption period). In addition, they show a significant difference in investment behaviour, operating performance and value creation. This result is consistent with the notion that an EPP-based compensation system encourages managers to choose profitable projects that ultimately maximize shareholders' wealth. However, they found significant differences between anticipated adopter firms and surprise adopter firms. The improved performance referred to above appears to have been driven mostly by anticipated adopters, which points to potential self-selection bias. The above studies looked at the adoption of EVA (or residual income) plans regardless of what measure was used before adoption. Balachandran (2006) argues that the original plan is important as adoption of EVA or residual income might actually increase or decrease investment pattern, even though the delivery of residual income would increase in both cases. This implies that firm value maximisation does not necessarily entail an increase or decrease in investments. Balachandran (2006) used a sample of 181 firms that adopted the residual income (RI)-based compensation incentive. These firms fell into two main categories: those which previously adopted earnings as a compensation plan and those which previously adopted return on investment (ROI) based compensation plan. He focused on two outputs, namely, the change in RI and the change in investment. His results show strong support for the view that RI-adopting firms do actually deliver higher RI after adoption. However, the results also show no significant change in investment pattern. Sample Our sample consists of 89 US firms that have adopted EVA as a compensation system. We follow Wallace (1997) and Kleiman (1999) and define the end of the first year of the company announcing its adoption of EVA as the event date ( = 0). The identity of adopting firms was found as follows. We began with Wallace's (1997) 23 firms that adopted the EVA compensation plan. We then added the firms used by Kleiman (1999). This increased the number to 71 firms. Subsequently, we searched for additional firms using various databases where the EVA adoption was specifically mentioned. These comprise the Stern Stewart & Co. brochure, Lexis-Nexis, Proxy Statement, 10-Q report and Wall Street Journal. The majority of firms which adopted EVA have disclosed such information in their official release. We identified a total of 101 adopting firms in the period 1987-2001. However, 12 EVA adopters were excluded because of lack of price/return and accounting data. Our final sample contains 89 EVA adopters on NASDAQ, NYSE and American Stock Exchange Markets. Because we use both the market model and control firm approaches to estimate abnormal performance, each of the adopter firm was matched with a benchmark firm. We follow Wallace (1997) who uses the Standard Industrial Classification (SIC) to match the adopter and benchmark firms. We used three criteria for matching firm selection. First, the company should have the same 4-digits SIC code (Same industry sector). If we do not find a firm with 4 digits match, we choose the best match with a 3-digits SIC code. If there are several firms with the same 4 or 3 digit SIC code, then we select the nearest size using a combination of the total asset and number of outstanding common shares in the year prior to the year of adoption to match adopters and control firms. Finally, if a benchmark firm does not have sufficient monthly data during the event period, we select the next best benchmark in terms of SIC code and size. 690 The price and return data of both adopting and control firms were collected from CRSP database. Appendix 1 provides a breakdown of firms adopting EVA and the year of adoption, the main control firms and the SIC code respectively. Method Defenders of Economic Value Added (EVA) claim that it helps to enhance the investment activity that leads to a notable market reaction (Stewart, 1991). The object this paper is to examine whether the adoption of EVA has the predicted positive effect on firm value. If EVA does indeed have such a positive effect on value, then the market should identify this change in the firm and react positively at least in the long run. If that is the case, then we should observe significant abnormal market performance (as measured by abnormal returns or abnormal buy and hold returns) either immediately (if the effect is true and the market is efficient), or slowly (if the market only absorbs information slowly). To test this claim we use an event study approach. Some event studies research has been dedicated to testing market efficiency. However, most of the event studies have been used to assess the impact of some events on some measure of the firm or investor wealth. Many studies have discussed and examined the long-term financial performances after the occurrences of certain events such as the IPO, mergers and acquisitions and the most popular event, cash dividends. One common feature of these studies is that of the classical event approach, which fully intended to investigate very short-term events. In a string of seminal papers, Barber & Lyon (1996, 1997 and Lyon, Barber, & Tsai (1999), revealed that the standard classical event study framework can lead to many partialities when applied to the measurement of long-term abnormal performances and recommended further study for such long-term events analysis. Further, Fama (1998) raised two important key issues regarding measuring long-term abnormal returns: first, the model's ability to correct for risk when estimating abnormal returns is quite low and second, the estimation of abnormal returns is probably subject to a range of statistical biases. Thus, to avoid estimation and bad model problems, we use direct benchmarking using both the market index and a matched control firm. The event window is set to 60 months before and after the adoption date. In the literature there are two methods to test the events and detect a long-run abnormal stock return: the cumulative abnormal return (hereafter, CAR) and the Buy and Hold Abnormal Return (hereafter, BHAR). The main difference between CAR and BHAR is mainly attributed to the compounding of the monthly return; while BHAR incorporates the effect of compounding CAR does not (Barber & Lyon, 1997). Regardless of the methodology used to measure the performance of the EVA adopter, CAR or BHAR, we need to measure the abnormal return. The abnormal return is the difference between the actual return and the benchmark return of a security. Events in the theory of finance can usually be classified as information that has not already been contained in the share's market price. Let be the return on adopter (event) firm , and be the return on a benchmark stock. When we use the marked index as benchmark, then we simply set = , where is the return on the market portfolio. Accordingly abnormal returns are given by = − The average abnormal return ( ) during month s can be defined as: where is the abnormal return estimator for security i and is the number of the companies in the sample during month t. The cumulative average abnormal return in the window of ( , ) is: = The second method used to calculate the abnormal return is buy and hold abnormal return (BHAR) which is defined as the compound returns on the event firm less the compound return on a control firm / reference portfolio-that is BHAR: is the return on the event firm (adopter firm) i in month t. is the benchmark returns. As our main method to test the event is BHAR it is more efficient to highlight the skewness problem inherited within the process of making inferences using BHAR. This problem is reported in Barber and Lyon (1997). In order to conduct the significance test in event time using BHAR, the following conventional t-statistic is used based on cross sectional data: is the sample mean, i  is the standard deviation, and N is the number of EVA adopter firms. The compound nature of BHAR induces skewness in the above statistic. To circumvent this problem, we use a bootstrap correction, originally proposed by Johnson (1978): Where ˆ is the coefficient of skewness, and / i S BHAR     . This adjustment was recommended by Lyon et al. (1999) to correct for potential skewness in BHAR returns. Kothari & Warner (1997) state that, drawing statistical inferences from a bootstrap approach is likely to be a better technique for statistical testing of long-term stock abnormal performance. However, while standard bootstrapping on the skewness-adjusted t-statistic does indeed address skewness concerns, it does not address the question of heteroscedasticity. In our case adoption takes place over a number of years. During these years there are periods where the market is highly volatile and others where it is relatively calm. Returns and, hence, abnormal returns, drawn from different periods are likely to have been drawn from distributions having different volatilities. The standard bootstrap does not correct for heteroscedasticity. The wild bootstrap, on the other hand, is designed to account for heteroscedasticity. Despite its properties, the wild bootstrap has unfortunately seen very little use in empirical finance. The standard bootstrap draws samples (with replacement) from the set of estimated variable. In our case, we draw samples This latter scheme is preferred since we use the skewness adjusted t-statistic which corrects for skewness, and thus skewness is not an issue (with the adjustment of Johnson (1978) the parent distribution of the adjusted statistic is expected to be symmetric). A further serious problem that we confront both in the EVA adopter sample and the control firm sample is that of firms that delist within the event period. Delisting can result from acquisition, bankruptcy or going private. Liu & Strong (2006) replace delisted firm returns by either zero or the risk-free rate. They find similar results in both cases. Lyon, Barber & Tsai (1999) and Mitchell & Stafford (2000, p.298) replace all de-listed firms with the benchmark return. This has the potential to create an upward bias in the estimated BHAR returns, since some of these de-listings are bankruptcies. However, for the purpose of our study we use the following rules. If an observation is missing within a valid set of observations, we set the return equal to zero. If the de-listings are due to bankruptcy, we replace the missing return by -1. Finally, if the delisting is due to a value preserving event such as a merger, we replace the return by the benchmark return. We use CRSP description as a distinguishing feature of the delisted firms. The Delisting Code is a 3-digit integer code. It either (1) indicates that a security is still trading or (2) provides a specific reason for delisting. All coded de-listings are categorized by the first digit of the delisting code. The second and third digits of the delisting codes provide further details of delisting events. Additional delisting codes, specific to various delisting categories, have been created to indicate when an issue is closed to further research, or if the issue is pending further research. The most important codes are 241, 231, 233, 331, 251, 552 and 574. These categories of delisting are most likely to be stocks that are either worthless or some distance from providing shareholders with any terminal value, and consequently we treat these cases as if investors lost all their investment. Cumulative Abnormal Return Results We begin by discussing cumulative abnormal returns. All results are displayed as follows. The average CAR, the standard tstatistics, the skewness adjusted t-statistic with its bootstrapped confidence intervals, the skewness and kurtosis of each set of cumulative abnormal returns is shown for selected months. The statistics are shown in a brief table that shows the 1 st , 10 th , 20 th and up to the 60 th month. Whenever the standard t-statistic is above 1.96, the results are also shown. The full 60 months' version of the table is provided in the appendix. To complete the picture, we provide a graphic summary of CAR, the skewness adjusted t-statistic and the 1% and 99% quantiles obtained from the wild bootstrap. Table 2 reveals an interesting story about the overall performance, the cumulative abnormal return (CAR) for the 10 years after the adoption date. Throughout the 5-year post-event period, most of the CARs appear to be positive but insignificant except for the months 27, 28, 29 and 30 which are in the third year after EVA adoption. During these months the mean CARs are at their highest levels and are significant. However, following these months, we find that CARs start to decline and then rebound up again around month 40. After that we see that CARs decrease and fall into negative territories. Especially in the last 4 months in year 5 this negative trend is quite visible indicating that EVA adopters underperform their matched non-EVA firms. Figure 1 depicts AR against CAR based on matching firms' benchmarking. Table 3 shows CARs based on market benchmarks (S&P 500), In nearly all months, adopter firms outperform the market with small variances which does not increase in the best of cases more than 8.5% at month 20 and this is the only significant month of the 5 year post-event period. The mean return based on CAR is positive for most months except for months 34, 35 and 36, which give negative CAR. All the CAR returns are skewed and leptokurtic, this might be attributed to the compounding process inherited in CAR calculations. However, the simple CAR based on matching firms and market benchmarking shows similar dynamics-the scale of the suggested outperformance is not the same. Firstly, the CAR based on matching firms is about twice as large as the CAR based on the market benchmark in the positive cases. Secondly, CAR based on benchmarking is only negative between months 34 and 36. This is followed by an apparent upward trend (Figure 2). Buy and Hold Return Results Our first results regarding the BHAR, presented in Table 4, show the BHAR derived using firm benchmark. The BHAR increases from an insignificant +1.0% after 9 months to a significant 36.4% after 27 months, becoming insignificant thereafter and continuing to increase to 37.4% after 41 months, 40.2% after 45 months, and then starts to decline to reach the lowest return of 4.3% after 59 months. All returns are skewed and leptokurtic. It is worth noting that the adopter's buy and hold return (BHAR) itself is highly skewed and leptokurtic throughout the period and that the matching firm is also skewed and leptokurtic, but to a lesser extent. Table 5 presents the results from a comparison with the benchmarking portfolio. Generally, the BHARs are smaller in value than those obtained with matching firms' benchmarks. Once again the BHAR has a positive and insignificant mean return through the hold period, it is positive at month one (3%), rising to the highest and insignificant mean return of 20.6% after 29 months, three years and one month after the event date. Beyond 29 months the rate of decline accelerates, with abnormal returns reaching 18.4% after 4 years and -2.7% after 5 years. The skewness and kurtosis of the BHAR based on matching firms is greater than under the benchmark S&P500. The difference being attributable to the new issue and rebalancing issue in the benchmark portfolio compared to the matching firm benchmark. There is a quite obvious difference between BHAR and CAR. CARs look more stable than BHAR when using matching firms to calculate the abnormality. On the other hand, the results based on the market benchmark are essentially the same and the produced curves are identical in most time periods. Even within BHAR itself. BHAR calculated using a matching firm appears greater than when using the benchmark market portfolio (S&P500). Overall, the behaviour of the aggregate abnormal return, CAR and BHAR, clearly appear to be sensitive to the method adopted to gauge the abnormality. Furthermore, BHAR based on matching firms grow faster than when based on the S&P500 benchmark especially after the adoption date where BHAR increased by more than 1.5 times. As discussed in the previous section the aggregate abnormal return, BHAR and CAR, is always highly skewed and leptokurtic and we suggest the wild bootstrapping (as discussed in methodology section) as a correction for these biases. This section will highlight the results of the bootstrap and the result of testing the null hypothesis that the aggregate abnormal returns, CAR and BHAR, are zero. The full version of the bootstrap test and tables are provided in Appendix (2). Fig. 3 depicts the skewness-adjusted t-statistic for the holding periods (60 months). The dotted and dashed lines are the 5 th and 95 th percentiles of the bootstrapped distribution. These can be interpreted at either the 5% critical value level for a one tail test, or the 10% critical value for a two-tail test. The graphs describe the two schemes of benchmarking: S&P500 portfolio and matching firms have a similar pattern but express different messages. The two graphs have the same feature which is that outperformance increases at around month 13 but there is then a slight variation with a different and insignificant range. For the BHAR-based market benchmark portfolio scheme the insignificancy remains hold throughout the holding period and the outperformance accelerates to reach the highest volume in month 20. Following this it is slightly volatile and reaches the lowest point of outperformance in month 39 after which it dramatically increases until month 45 where it then appears stable to the end of the holding period. Similarly, the BHAR based on matching firms copies its counterpart but the outperformance ceases from being significant at around the 25 -31 month and 37-42 month period. However, the aggregate BH return rapidly decreases after month 47 to reach close to zero as shown in Figure 4. The CAR based on the matching benchmark provides a different story: the graph in Figure 4 shows that CAR behaviour becomes more erratic and is no longer significant beginning from around the period 25-31 months. The performance of the adopting firms is quite low, almost zero after the adoption date and sometimes underperforms the matching firms as depicted in Fig. 4. Discussion In general, the purpose of this paper is to investigate whether the adoption of EVA as a compensation and management plan will positively affect the performance of adopting companies. The paper compares the performance of adopting firms to that of selected matching firms and to the market indexes particularly the S&P500 portfolio. Then it uses two common aggregating methods to test the event of adopting EVA by different US firms namely the CAR and BHAR methods. The results obtained however, showed a slight improvement in the performance of companies adopting EVA within five years from the date of adoption. This is implicitly in line with what Wallace (1997) concludes in this regard. Wallace indicated that adopting EVA will encourage managers to take decisions that will lead to efficiently using the firm's assets to increase the wealth of shareholders and the value of firms through taking accurate decisions regarding the investing, financing and operating activities. This, in turn, will be reflected in the price of shares in the stock market, therefore improving the performances of these stocks. Similarly, the results achieved is incompatible with that of Kleiman (1999) where he compares the performance of firms adopting EVA to the performance two set of matching firms, the industry peer and closest match peer. By comparing the median of abnormal return he found that EVA adopter's show better performance after the adoption and outperform both the industry peer and closest peer match firms. The adjusted market return increases from 2.8% to 28.8% through three year time period after the adoption for the industry peer and from 2.6% to 7.8% for the closest match peer. However, the increases in performance of the adopting firms are still quite low. I used the mean of CAR and BHAR to compare the performances of adopting firms to those matching firms and market benchmark portfolio (S&P500 index) and the result revealed that EVA's firms outperform those matching and S&P500 portfolio and the CAR increases to reach 8.85% and 36.6% for matching firms and benchmark index respectively and the BHAR increases to 6.6% and 26.8% for the same order. Conclusion This paper has described the research design and the methodology that was used to examine the EVA adoption event. Both the CAR and the BHAR approaches were adopted to conduct our study. The previous research has been extended by increasing the number of EVA adopter firms to 89 and the time horizon of the study to cover the firms' performance during the period 1960-2012 was also extended. In addition, wild bootstrapping and using the skewness adjusted t-statistic to enhance the statistical reliability of the event test statistics was adopted. By doing this all three moments of the parent distribution of the test statistic (heteroscedasticity, skewness and kurtosis) were taken into account. Furthermore, the criterion to select the matching firms was carefully applied as was the problem of delisting. The results obtained in this research are consistent with the previous studies' results discussed in section 5.2. Regardless of the methodology approach, CAR or BHAR, the results of this chapter reveal that firms adopting EVA as a compensation plan and management tool outperform the market (S&P500) and matching firms (same sector) most of the time within the hold period. The CAR results show that despite the benchmarking used the majority of adopter firms positively outperform the matching firms and the S&P500 portfolio and for a few months the adopter firms have a negative performance mainly in year one and year five of the 10 year estimated period. In general, CAR appears more stable and has the lowest skewed and leptokurtic. Regarding the BHAR approach the findings reveal that the mean return of the adopter firms is both positive and highly skewed and leptokurtic throughout the holding period. Generally, the results obtained from a comparison against the benchmarking portfolio (S&P500) are smaller in value than those obtained when compared to the matching firms' benchmark. One interesting finding is that CAR is almost the same as BHAR when the S&P500 portfolio is used as a benchmark to calculate the aggregate returns. To sum up, irrespective of the aggregation approach used to measure the abnormal return, the adopter firms have a considerably low outperformance and this outperformance increased as the hold period increased. However, even with the positive performances most EVA adopter firms' outperformance declines after the adoption and takes some time to return to negative performance when matching benchmarks are used. This might typically reflect the fact that the market might react poorly to the adoption announcement. Finally, by analysing the adopter firms' performance we recognize that the adoption exists after a period of bad performances.	2020-07-09T09:03:27.725Z	2020-01-01T00:00:00.000	{ "year": 2020, "sha1": "d386d398d3c8af54ffb9494680c972a9151b42d2", "oa_license": "CCBY", "oa_url": "https://doi.org/10.5267/j.ac.2020.6.015", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "5d338e6d2d9fe788da47dc8cb6d801e80789a0df", "s2fieldsofstudy": [ "Economics", "Business" ], "extfieldsofstudy": [ "Economics" ] }
42520499	pes2o/s2orc	v3-fos-license	Spontaneous retroperitoneal hemorrhage presenting as hemoperitoneum secondary to renal cyst rupture in a peritoneal dialysis patient with acquired cystic kidney disease Spontaneous retroperitoneal hemorrhage (SRH) is a rare and potentially fatal condition. Acquired cystic kidney disease (ACKD) may cause SRH in hemodialysis patients. However, presentation of retroperitoneal hematoma as hemoperitoneum in peritoneal dialysis (PD) patients is exceedingly rare. We report a 44-year-old male PD patient who presented with hemoperitoneum secondary to retroperitoneal hematoma. The reason of SRH was rupture of the cysts of ACKD. The patient underwent unilateral nephrectomy with subsequent disappearance of hemoperitoneum. The importance of this case lies in the fact that the patients who have been receiving dialysis for a long time should be under surveillance in terms of ACKD development and potential associated complications such as cyst hemorrhage and malignancy. into the perinephric space leading to perirenal hematoma, or retroperitoneal hemorrhage (RH).Perinephric hematomas have been reported in up to 13% of patients with ACKD. [1,3]moperitoneum is an unusual complication, and may present in up to 8.4% of cases in patients undergoing peritoneal dialysis (PD). [4,5]To the best of our knowledge, hemoperitoneum, as a complication of renal cyst rupture in autosomal dominant polycystic KD (ADPKD), has been described only twice in patients on PD. [6,7] On the other hand, there is only one reported case, in which hemoperitoneum was secondary to cyst rupture of ACKD in PD. [8] Presentation of retroperitoneal hematoma as hemoperitoneum is exceedingly rare. We describe a PD patient with ACKD who developed hemoperitoneum secondary to retroperitoneal hematoma along with a discussion of the relevant literature. Case Report A 44-year-old man with ESRD caused by hypertensive nephropathy had been treated with hemodialysis for Introduction Acquired cystic kidney disease (ACKD) occurs in patients who are on dialysis for end-stage renal disease (ESRD).The prevalence of ACKD is directly related to the duration of dialysis. [1]Hemorrhagic renal cysts are the most frequent complication of patients with ACKD. [2]Bleeding is usually confined within the cyst but occasionally extends into the renal collecting system leading to hematuria or 3 years before being converted to PD.The patient presented with abdominal pain, left-sided loin pain and bloody peritoneal effluent during the 7 th year of PD treatment.His blood pressure was 140/80 mmHg and heart rate was 82 bpm and rhythmic.He had low grade fever of 38.3°C.There was tenderness over abdomen and left loin.There was neither rigidity nor rebound tenderness over these areas.He denied diarrhea, vomiting, syncopal episode, and bloody stool.His peritoneal effluent seemed hemorrhagic, presented with red blood cells of 2000/mm 3 and 0.1% of hematocrit.White cell count and differential of the peritoneal outflow fluid revealed a total count of 1540/mm 3 (83% neutrophils).Laboratory values were as follows; white blood cells: 16.4 × 10 3 , hemoglobin: 8 g/dl, C-reactive protein: 69 mg/L, erythrocyte sedimentation rate: 46 mm/h, and procalcitonin: 14.2 mg/dl.Coagulation profile and liver tests were within normal levels.Blood and peritoneal fluid were sent for culture and the patient was placed on ceftriaxone and vancomycin treatment with a presumptive diagnosis of continuous ambulatory PD-related peritonitis.Blood and peritoneal fluid cultures were negative for any bacterial and fungal infection.The hemoglobin value of the patient 2 days prior to current admission was 12.4 g/dl.The patient denied any rectal bleed or bloody vomiting.PD was halted and the patient underwent heparin-free hemodialysis with saline flushes.Abdominal ultrasound showed bilateral shrinked kidneys with multiple small-sized cysts.There was an exophytic cyst measuring 9 cm × 8 cm, filled with blood at the lower pole of the left kidney.Medical records showed that he had no cysts at the start of the renal replacement therapy.Abdominal computed tomography revealed that there was a lesion, which extends from the upper pole of the left kidney to the lower pole through which a hematoma was filling renal parenchyma, subcapsular space and perirenal area [Figure 1].The patient was transfused with packed red blood cell suspensions and underwent left radical nephrectomy with presumptive diagnoses of renal cell cancer secondary to ACKD or cystic hematoma.Macroscopically, there was a cystic lesion, which was not demarcated from surrounding tissues and filled completely with necrotic bleeding tissues.Histopathologically, no focus of neoplasia was found [Figure 2].After nephrectomy, hemoperitoneum disappeared and the patient recovered completely.He is still undergoing hemodialysis thrice weekly. Discussion The usual definition of ACKD requires three or more cysts per kidney in a patient on dialysis who does not have a hereditary cause of cystic disease such as ADPKD or tuberous sclerosis.Within the first 3 years of dialysis, approximately 10-20% of patients develop ACKD.By 5 years, 40-60% of patients have ACKD and by 10 years more than 90% of patients exhibit ACKD. [9]There is no difference in prevalence and severity of ACKD between hemodialysis and PD patients. [3]r patient had risk factors for development of ACKD, including male gender and long duration of renal replacement therapy.The patient had no family history of ADPKD; moreover, he did not have any cysts in the kidney at the start of the dialysis.Nephrectomy material also histologically confirmed the diagnosis of ACKD [Figure 2].Approximately 50% of patients with ACKD develop hemorrhagic renal cysts. [1,2]On the other hand, this rate is much lower in patients with ADPKD (18% in a cohort of 158 patients during a 7-year follow-up). [10]Spontaneous retroperitoneal hemorrhage (SRH) is rare but potentially life-threatening condition. [11]Causes of SRH includes myriad etiologies but the most common cause is kidney neoplasm, angiomyolipoma being predominant. [12]The frequency of SRH is not known in patients undergoing renal replacement therapies, because many studies excluded these patients.Malek-Marín et al. [13] reported a single center incidence of SRH in hemodialysis patients as 0.86 cases/100 patients.When SRH was of renal origin in dialysis patients, the most common cause was cyst rupture in patients with ACKD. [14]st patients with SRH secondary to cyst rupture were undergoing hemodialysis compared with PD.This observation has been explained by longer duration of RRT and consequently more severe cystic disease, and heparin anticoagulation. [1]other notable observation in SRH secondary to cyst rupture is that there is a remarkable difference in frequencies between SRH due to ACKD and ADPKD.Most of the described patients in the literature had underlying ACKD rather than ADPKD.Some putative factors have been put forward to explain this disparity.The more medullary location of cysts in ADPKD compared with cortically residing cysts of ACKD and differences in the speed of cyst growth might account for differences in observed prevalence of SRH. [13]Among hemodialysis patients, perirenal or retroperitoneal bleeding due to acquired cyst rupture tends to occur after the dialysis session, possibly as a consequence of heparinization. [3]e of the interesting aspects of the present case is the manifesting as hemoperitoneum instead of RH.A review of the causes of hemoperitoneum by Nace et al. [15] mentioned mostly intraperitoneal pathologies although it does include some extraperitoneal sources.Peritoneal dialysis is a window to the peritoneum and is thus a sensitizer revealing peritoneal pathologies which might otherwise go undiagnosed. [16]To our knowledge, hemoperitoneum, as a complication of renal cyst rupture in ADPKD, has been described only twice in patients on PD.On the other hand, hemoperitoneum secondary to cysts rupture in a PD patient has only been reported once to date. [8]This is not surprising because of retroperitoneal hemorrhage, in theory, should not find its way through the peritoneal membrane.A possible explanation is an adhesion between the cyst's wall and the peritoneum, favored by their anatomical proximity and inflammation secondary to intracystic hemorrhage.These adjoining structures could then rupture as a result of the rising intracystic pressure.Among the more common causes of hemoperitoneum, SRH should also be kept in mind. Since most of the clinically overt SRH cases in dialysis patients are secondary to rupture of acquired cysts, especially after 5 years of renal replacement therapy, patients should be screened for the presence of ACKD.In patients who had ACKD, this surveillance should be continued because of risk of cyst enlargement and malignancy risk. Conclusion Our case report exemplifies a rare PD patient who developed RH presented as hemoperitoneum.The reason of SRH was rupture of cysts of ACKD.SRH is an important and potentially fatal complication.Patients who have a long duration on dialysis should be under surveillance related to ACKD development and associated bleeding risks.	2018-04-03T05:23:27.497Z	2014-12-01T00:00:00.000	{ "year": 2014, "sha1": "76c5fe085512e365a2109fab11ee8cdf63e427f9", "oa_license": "CCBYNCSA", "oa_url": "https://doi.org/10.4103/0971-4065.147371", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "aeef6753f25f2eecf8a3aec3ba30ccc444769952", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
49183614	pes2o/s2orc	v3-fos-license	Tuberculous Arthritis of the Elbow Joint: An Uncommon Location with a Diagnostic Dilemma Musculoskeletal tuberculosis accounts for 1%-3% of all cases of tuberculosis (TB) worldwide with elbow involvement being even less common. The most cases of tuberculous arthritis occur in patients born in and emigrated from endemic regions, especially in patients who are co-infected with human immunodeficiency virus (HIV). We present a rare case of tuberculous septic arthritis of the elbow joint in a 78-year-old African-American female from the United States, with no history of travel abroad. Her presenting symptoms included pain, swelling, and decreased range of motion of the right elbow for six months. She underwent incision and debridement of the elbow joint and was started on empiric intravenous antibiotic therapy for suspected pyogenic septic arthritis. Several weeks later, surgical cultures demonstrated acid-fast bacilli, identified as Mycobacterium tuberculosis (M. tuberculosis) and a four-drug anti-tuberculosis regimen was initiated. Based upon culture results, additional imaging evaluation was undertaken. She did not have any symptoms of a pulmonary disease but was found to be positive for Mycobacterium tuberculosis in sputum cultures and bronchoalveolar lavage. We emphasize the importance of considering a tuberculosis infection in the differential diagnosis of monoarticular arthritis, especially in elderly patients with immune deficient states since early recognition and treatment result in good functional outcomes. Introduction Infections due to Mycobacterium tuberculosis (M. tuberculosis) occur worldwide, with humans being the only known reservoir. Of the estimated two to three billion people infected with M. tuberculosis, only a relatively small percentage (5%-15%) of people will develop clinical disease [1]. Lungs are the most common organ of involvement. Extrapulmonary tuberculosis (TB) makes up about 15%-20% of all cases of TB in immunocompetent hosts and 50% in those individuals who are affected with human immunodeficiency virus (HIV) [2]. Musculoskeletal TB accounts for 35% of extrapulmonary TB cases [3] and 1%-3% of all TB cases worldwide [4][5]. The most commonly infected joints are the spine, hip, and knee in order of frequency [5]. An infection of the elbow joint is particularly rare, making up 1%-5% of all cases of musculoskeletal TB [6]. We are presenting a rare case of a 78-year-old female patient who presented with a presumptive diagnosis of pyogenic septic arthritis, which was later unmasked as tuberculous septic arthritis of the elbow. Case Presentation A 78-year-old African-American female presented to the hospital with complaints of worsening right elbow pain and swelling for six months. She described the pain as a dull ache that worsened with movement and, over time, was unable to fully move the arm. She denied any history of trauma or any other joint involvement. She had a past medical history of stage IIIC ovarian Mullerian tumor with peritoneal carcinomatosis, requiring abdominal hysterectomy with bilateral salpingo-oophorectomy and neoadjuvant chemotherapy with paclitaxel and carboplatin, which was completed two years ago. She was born in the United States and was a retired accountant. She denied any travel outside the country or exposure to TB cases in the United States. Her medications included meloxicam and alendronate. On initial presentation, her vitals were significant for a temperature of 102.6 F, a respiratory rate of 22 breaths/minute, a pulse rate of 87/minute, and blood pressure of 124/66 mm Hg. The physical examination was remarkable for right elbow erythema, swelling, effusion, and tenderness to palpation. Her range of motion was restricted to only five degrees of flexion and extension at the right elbow joint. The laboratory evaluation revealed leukocytosis of 12.3 x 103/µL, a uric acid level of 7.1 mg/dL, a C-reactive protein of 115 mg/L, and an erythrocyte sedimentation rate of 104 mm/h. An X-ray and computed tomography (CT) of the elbow joint demonstrated bony erosive changes, joint space narrowing, and large effusion in the elbow joint. Magnetic resonance imaging (MRI) of the elbow joint revealed large joint effusion, synovitis, and osteomyelitis of the surrounding bones ( Figure 1). Blood cultures were drawn and the patient was started on intravenous vancomycin and piperacillin-tazobactam for presumptive pyogenic septic arthritis. She underwent incision and drainage of the right elbow after an unsuccessful attempt at arthrocentesis. Preliminary blood and surgical cultures were negative for bacterial growth. She was discharged home with empiric intravenous antibiotic therapy for six weeks, home wound care, and occupational therapy. The patient's symptoms initially improved but about three weeks later, she returned to the hospital with worsening right elbow pain and swelling. Acid-fast bacillus (AFB) cultures sent at the time of initial incision and drainage became positive for acid-fast bacilli (Figure 2), and her antibiotics were switched to azithromycin, meropenem, and doxycycline to cover for atypical mycobacteria. A deoxyribonucleic acid (DNA) probe of the acid-fast bacilli was positive for Mycobacterium tuberculosis (MTB), and the patient was started on four-drug antituberculosis therapy, including rifampin, ethambutol, isoniazid, and pyrazinamide with pyridoxine. Final culture results confirmed the presence of the Mycobacterium tuberculosis complex and susceptibilities revealed a non-resistant MTB strain. She was then evaluated for disseminated disease with a CT chest, abdomen, and pelvis. Her CT chest results were remarkable for multiple left, upper lobe nodules and left lower lobe air-space consolidation ( Figure 3). The patient denied ever having any symptoms of cough, shortness of breath, hemoptysis, and weight loss. She did report having night sweats and chronic fatigue. Bronchoscopy was performed and bronchoalveolar lavage fluid cultures were positive for MTB. She was diagnosed with active TB of the lungs and the elbow joint, possibly as a reactivation of latent TB. The HIV test was negative. The patient was discharged home on the four-drug antituberculous regimen for 12 months under the supervision of the health department. At a fourmonth follow-up visit, the patient showed significant improvement in her symptoms. Discussion An estimated 2-3 billion persons worldwide have latent TB infection and only 5%-15 % develop an active infection during their lifetime [1]. Lungs are the most common site of involvement with active TB. About 25% have asymptomatic pulmonary infection [7]. Over the past six decades, TB case rates have been consistently trending down; however, susceptible groups remain at high risk in the United States (U.S.) African-Americans represent the highest racial group, constituting 37% of TB cases among the U.S.-born persons [8]. Predisposing factors for TB in the United States include birth in TB endemic areas and recent immigration, poor socioeconomic conditions, injection and non-injection substance abuse, diabetes, hematological malignancy, head and neck solid organ malignancy, aging, and immunodeficiency, including HIV, solid organ transplant, and tumor necrosis factor-alpha inhibitors [2, [9][10]. Musculoskeletal involvement occurs from a dormant pulmonary or extra-osseous focus by a hematogenous, lymphatic, or direct spread of bacilli [6]. Tuberculosis of the elbow usually starts with olecranon or the lower end of the humerus, but in some cases, the primary lesion involves the synovium or the upper end of the radius [11]. The tuberculous lesion is almost always a combination of osteomyelitis and arthritis. Initially, an inflammatory reaction develops in the synovium followed by granulation tissue formation, progressing to the development of effusion with fibrin deposition, forming "rice" bodies. This pannus of granulation tissue then begins to destroy cartilage, leading to bone demineralization and caseous necrosis. Cartilage is destroyed peripherally, preserving joint space for a considerable period of time, which has important clinical implications [12]. Tuberculous arthritis usually presents insidiously with chronic joint pain accompanied by swelling, progressive loss of function, and local muscle wasting. Constitutional symptoms, including low-grade fever, weight loss, lassitude, night sweats, anemia, and tachycardia, are present in a significant minority of cases [4]. In approximately 50% of the cases, symptoms and radiographic evidence of pulmonary TB are absent and tuberculosis is frequently missed as a differential diagnosis of the chronic inflammation of joints in the absence of active pulmonary disease. The differential diagnoses of skeletal TB include pyogenic arthritis, rheumatoid arthritis, sarcoid arthritis, gout, pigmented villonodular synovitis, and tumors [6]. Early diagnosis of skeletal TB is essential since the preservation of joint space early in the disease leads to good functional outcomes if the condition is treated in the early stages. A tuberculin skin test, usually positive in musculoskeletal TB, is one of the most valuable diagnostic tests [12]. Confirmation by microscopy and culture is necessary, and the joint fluid aspirate should be evaluated for smear and culture, as it yields positive culture results in 80% of the cases [3,12]. A clear synovial aspirate is an excellent specimen for the polymerase chain reaction and nucleic acid probes [4]. On a plain radiograph, a Phemister triad suggesting tuberculosis arthritis may be observed in some cases. This triad includes juxta-articular osteopenia, peripherally located osseous erosions, and a gradual narrowing of the joint space [5]. Martini et al. divided the radiological presentation of osteoarticular tuberculosis into four stages: stage 1 -localized osteoporosis, but no bony lesion; stage 2 -one or more erosions or cavities in bones; stage 3 -involvement of the whole joint without gross destruction; stage 4gross destruction [13]. MRI and CT scans can further aid in diagnosis, especially in determining the extent of the disease. A definitive diagnosis can be established by obtaining a biopsy specimen from diseased tissue, synovium, or bone and demonstrating positive cultures for Mycobacterium tuberculosis and caseating granulomas [3,4,6]. The cornerstone of treatment for musculoskeletal TB is three to four drug regimens based on sensitivity results. The optimal duration of treatment is uncertain with the conventional duration being 12-18 months. However, recent data suggest that the duration of six to nine months of therapy can be appropriate in regimens containing rifampin [2]. Good functional recovery can be obtained with prompt diagnosis and treatment with anti-tuberculous chemotherapy and early mobilization. Surgery is rarely necessary in the modern era since the advent of chemotherapy. Advanced cases of peripheral joint involvement can be successfully treated with synovectomy, debridement of necrotic tissue, and abscess drainage without needing arthrodesis. With effective chemotherapy, surgery should be reserved only to prevent deformities and improve function when the disease has not responded to chemotherapy well [12]. Conclusions Tuberculosis of the elbow joint, in particular, is a rare entity and very few cases have been reported in the United States. Our case is unique because the patient did not have any exposure or significant risk factors for TB. Her presentation of refractory elbow pain led to further workup and an eventual diagnosis of tuberculous arthritis of the elbow as well as concurrent pulmonary tuberculosis, which was otherwise asymptomatic. Our case serves as a reminder to consider osteoarticular tuberculosis in the differential diagnosis of chronic monoarticular joint pain and swelling even if significant risk factors are absent. We stress the importance of early diagnosis and treatment to prevent the rapid progression of the disease and poor functional recovery at late stages.	2018-06-16T00:30:07.758Z	2018-04-01T00:00:00.000	{ "year": 2018, "sha1": "0e3625547be1c68afa2aa2aa60db1cf436249005", "oa_license": "CCBY", "oa_url": "https://assets.cureus.com/uploads/case_report/pdf/11582/1528830587-20180612-4885-rd3oxj.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "5db91511e3bcab16a11d78a98d34149342cf49f8", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
56084728	pes2o/s2orc	v3-fos-license	Towards Comprehensive Policy Integration for the Sustainability of Small Islands : A Landscape-Scale Planning Approach for the Gal á pagos Islands Accomplishing and implementing sustainable development goals in the context of insular socio-ecological systems requires effective policy integration—i.e., the integration of policy actors across multiple sectors and levels of government to improve policy outcomes. However, achieving policy integration entails significant challenges because it is highly context-dependent. This study investigates policy integration within the complex socio-ecological context of the Galápagos Islands in Ecuador. The paper analyses Galápagos legal and planning documents to evaluate the extent to which they support comprehensive policy integration. The analysis found that recently adopted institutional arrangements have strengthened government institutions at the provincial level, and started to consider concepts relating to socio-ecological and land–sea management. Nevertheless, key policy actors and pressing issues remain unattended, due to policy inconsistencies, institutional arrangements limitations, and fragmented approaches to conservation and development control between provincial and local governments. Insights are presented to improve the comprehensiveness of policy integration in Galápagos based on a landscape-scale planning approach. Introduction This paper discusses the need for more effective policy integration to accomplish sustainable goals within the context of small islands' landscapes.Small islands-populated and/or unpopulated islands without both full political autonomy and independence [1]-are of paramount importance for socio-ecological wellbeing because they host countless interdependent land-marine ecosystems of outstanding landscape qualities, and of highly rare and endemic biodiversity [2][3][4][5].The one-off environment of every small island is the result of the delicately balanced, complex, constantly evolving, and mutually modifying processes between the relatively few living organisms that attempt to colonize it and its unique geophysical characteristics (e.g., location, geographical origin, degree of isolation, availability of resources, and scale) [2,3]. In parallel, while humans inhabiting small islands have adapted to their conditions and influenced the generation of complex socio-ecological systems, they are their primary active drivers of change [6][7][8]. In particular, the highly-specialized islands' socio-ecological systems seem unable to cope with the rapid changes generated by ever intensifying and conflicting land-sea uses (e.g., tourism, fishing, agriculture, urban development) [1,[9][10][11].These generate dynamic and often unforeseen consequences that could potentially result in severe environmental issues, such as the introduction of invasive alien species, habitat destruction, pollution, and the depletion of natural resources [6,12].Subsequently, these could drive the loss of islands' native biodiversity, jeopardizing the long-term provisioning of ecosystems services upon which human wellbeing depends [4,11,13]. Additionally, small islands' local governments and communities have minimum to no influence over many of the drivers of land-sea uses, such as the political agendas of higher levels of governance, and the globalized market demands for job opportunities, tourism destination, and fishing products [1,14].Small islands' socio-ecological systems are also under threat from natural hazards and the worsening impacts of human-driven climate change [11,14].Furthermore, small islands' remoteness, unstable political landscape, limited institutional and organizational capacity, and insufficient economic resources (local, national and international) constitute major obstacles for addressing the cross-cutting issues affecting them and for receiving much needed technical and financial assistance [1,11,[14][15][16]. To ensure the sustainability of islands' socio-ecological systems, multiple international policy frameworks have developed a broad range of goals and tools.Several of these frameworks have focused on increasing the protection status and improving the sustainable management of islands' social, cultural, and ecological iconic features (See [17][18][19][20]).Others have opted to promote the sustainable development of small islands developing states (i.e., 37 United Nations' members and 20 associated island states) (See [21][22][23]).These endeavors have led to the formulation of multinational to subnational institutional arrangements for the protection of many World Heritage Sites [24], biodiversity hotspots [20], and marine protected areas [25].Additionally, they have been instrumental in the preparation of several international initiatives for sustainable development (see [26][27][28]). Nevertheless, these frameworks remain tainted by significant shortfalls that diminish their capacity to trigger the institutional, organizational, and cultural change that is necessary to deal with the complex, rapid, and ever-changing island contexts [28][29][30][31][32].For instance, the poor legally-binding nature, gaps, and duplication of these arrangements, limit their enforcement and implementation, particularly at national and subnational levels [28][29][30][31].More importantly, the success or failure of policy measures is ultimately context-dependent.This means that policies are facilitated or inhibited by a raft of national and subnational political factors (e.g., institutional, organizational, financial, management, socio-cultural, environmental) that need to be appropriately considered [33,34].Additionally, divided approaches within and between social and ecological sciences, science and policy sectors, government and non-government actors, and local/traditional and technical knowledge have constrained the generation and sharing of knowledge that informs policymaking processes [1,11,28,[35][36][37][38]. Such compartmentalized approaches result in a poor understanding of the temporal and spatial implications [35,36] that policies may inflict on socio-ecological [39,40] and land-marine [41,42] systems. Accomplishing sustainable development goals in socio-ecological systems requires the timely and collaborative work of a broad range of policy actors (e.g., individuals, social groups, governments, non-government organizations) with a stake and knowledge on the matter across multiple sectors and levels of governance.These must come together to recognize interdependencies between sectors, solve sectoral disputes, define common short-to long-term goals, and determine the most appropriate way to achieve them [33,43].To this end, both ecological and social theory have advocated for more effective policy integration as a suitable means to raise awareness about cross-sectoral interdependency, build and expand links between sectors and levels of governance, and improve policy outcomes through the vertical and horizontal articulation of policy frameworks [33,35,36,[43][44][45][46].Thus, policy integration necessitates the comprehensive consideration of interdependent policy actors and issues, and the spatial and temporal dimensions through which these develop across [35,36].Planning at the landscape scale has emerged as a potentially suitable approach to both deal with the dynamic temporal and spatial dimensions of places and bring together multiple interrelated sectors through participative planning and policymaking processes [38,41,[47][48][49][50].In addition to all these considerations, achieving sustainable development goals in small islands requires balancing the protection of the broad range of threatened socio-ecological features that define each small island's insularity (e.g., biodiversity, identity, coastal/sea lifestyle) with the immense input of technical and financial resources that these require to accomplish sustainable goals [1,51,52].In this sense, the main differences with other socio-ecological systems are the small room for error that small islands leave to planning, due to their fragile functioning, unique conditions, and the latent anthropogenic and natural threats these face [11,14]. This paper brings together the concepts of landscape-scale planning, small islands' sustainability and policy integration to advancing the conservation of socio-ecological systems in the context of the Galápagos Islands.Applying the landscape-scale planning approach, the paper aims to glean insights to contribute to the improvement of the Galápagos' institutional arrangements (expressed through the legal and planning frameworks) so that these are better positioned to accomplish sustainability-related goals. Galápagos Islands offer a suitable case study because it represents a large number of small islands that are increasingly gaining more political autonomy.Achieving sustainable development goals through policy integration in this complex social-ecological system requires balancing the conservation of Galápagos' unique ecosystem with the pressures that population growth, fishing, tourism, and all anthropogenic activities associated with these, exercised over the limited natural resources.Moreover, in contrast with the large scholarly attention given to the ecological and biological aspects of Galápagos, literature addressing socio-ecological and policy issues in this archipelago is relatively scarce.There is a small number of studies that focus on very specific issues, such as the governance/management of the Galápagos Marine Reserve [53][54][55], environmental conservation [53,56,57], waste management [57,58], water supply [59,60], tourism [61,62], fishing [63], and socio-ecological interactions [64][65][66][67].Furthermore, these present a well-defined provincial scope.As a result, interdependent issues are partially analyzed, and little to no attention is given to broader national context and to the links and differences between the three Galápagos' municipalities.This study contributes to filling these gaps in knowledge by investigating the institutional arrangements from international, national to local levels, and assessing the role these arrangements play in Galápagos' provincial and local policymaking. To this end, the paper is structured in four parts.The first part includes a literature review on social and ecological theory focused on policy integration and island sustainability-related issues.Based on this literature review, an analytical framework is then proposed around four commonly agreed parameters for effective policy integration to achieving improved conservation in the context of small islands' socio-ecological systems.The second part describes the research approach, including case study description and data collection and analysis.Applying the analytical framework developed in part one, the third part presents results.Lastly, the forth part discusses and provides insights to advancing policy integration for sustainability of socio-ecological systems based on a landscape-scale planning approach. Materials and Methods The analytical framework applied in this study brings together the concepts of policy integration, sustainability and landscape-scale planning.The framework was used to carry out a document analysis of international, national, and regional legal and planning documents to distil insights for facilitating the accomplishment of sustainable development goals in the Galápagos Islands through effective policy integration facilitated by a landscape-scale planning approach. Unpacking the Concept of Policy Integration There is a large body of literature that attempts to define the concept of policy integration and related terms.For example, Underdal's [35] seminal work defined policy integration as the "extent to which a policy recognizes its consequences as decision premises, aggregates them into an overall evaluation, and penetrates all policy levels and all government agencies involved in its execution" (p.162).More recently, Shannon and Schmidt [43] described policy integration as an activity to facilitate, strengthen, and expand links between policy actors, organizations, and networks across sectoral boundaries.Related terms, such as policy coordination [68,69] and policy coherence [70,71], refer to the collaborative work between private and public actors to formulate policies that avoid gaps, duplication, or contradiction.Others, such as intersectoral cooperation [72] or cross-sectoral policy cooperation [73], indicate the collaboration between government and non-government policy actors through activities that assist in the identification of issues, definition of goals, and the implementation of policies and programs. In parallel, Stead and Meijers [33] provided a useful differentiation between the concepts of policy cooperation, coordination, and integration.Based on inputs and outputs to the policy cycle, Stead and Meijers arranged these concepts as consecutive stages that respectively denote higher levels of collaboration in policymaking.In the first stage, policy cooperation implies the collaborative work between policy actors working within their respective sectoral boundaries and following sectoral goals to produce more efficient sectoral policies [33].In the second stage, policy coordination requires the definition of common goals to address cross-cutting issues.Although the outcome remains to be more efficient sectoral policies, these policies are adjusted to be mutually enforcing and consistent to each other [33].Lastly, in the third stage, policy integration entails transcending sectoral boundaries for the definition of common goals to address cross-cutting issues through joint or cross-sectoral policies [33].Furthermore, the characteristics of these consecutive stages present great similarities to those of multidisciplinary, interdisciplinary, and transdisciplinary scientific research [74,75]. A range of positive outcomes usually associated with policy integration are relevant to small islands context.For example, policy integration endeavors may facilitate the formulation of articulated and/or joint policies to address cross-cutting issues [33,35,43].It may also help to split the cost of policy formulation and implementation among policy actors [33,35], and increase compliance with legal and policy frameworks [76].Additionally, policy integration processes may lead to progressively improving the capacity of policy actors to deal with more complex issues, trade-offs and uncertainty [70,77].This may improve the resilience of socio-ecological systems [77-79], and help develop local stewardship for the implementation, monitoring, and evaluation of policy measures [29,53]. Nevertheless, achieving effective policy integration is not a straightforward process.In particular, formulating articulated and/or joint policies requires greater commitment and more resources (time, effort, funding) from policy actors [33,35,43].Policy integration demands a more comprehensive understanding and assessment of the issues that trigger the policy cycle, and of the cross-sectoral consequences of policy decisions [35,36].More importantly, because policy integration implies shifting sectoral aspirations towards the common good, it entails some degree of cultural change that might be perceived as a loss of decision-making power, autonomy, and identity [32,33].Overcoming such perceptions requires offering stakeholders the prospect of short-to long-term benefits, incentives, and compensations [33,35].However, the benefits of policy integration initiatives often unfold in the mid-to long-term, and their initial costs might only be accepted when their benefits clearly outweigh short-term losses [35,36].Furthermore, policy measures and their respective outcomes are unlikely to be replicated, or succeed, outside their original socio-political context [33,70]. The initial and perhaps most direct path to trigger effective policy integration is to robustly define principles, guidelines, and policy goals, and include these into the legal framework to be followed, to enable compliance by all government agencies or policy actors involved [35,36].Although legal and institutional measures might change certain procedures, long-lasting cultural change can only occur when these changes are internalized through the first-hand experience and ownership/stewardship of policy actors over policymaking processes [35].This requires creating arenas for stakeholders' dialogue, knowledge and information sharing, and conflict resolution [35,36].Consequently, it is crucial to have flexible institutional arrangements to allow changes to occur in the policymaking process in the light of new information and lessons [35,36]. Analytical Framework: Landscape-Scale Planning, Small Islands Sustainability, and Policy Integration The analytical framework underpinning this paper is based on the synergistic relationship between the concepts of landscape-scale planning, sustainability, and policy integration.Planning at the landscape scale has emerged as a potentially suitable approach to address the wicked issues confronted by socio-ecological systems, including their long-term sustainability and resilience [80].In particular, this planning approach recognizes landscapes as being the result of dynamic interactions between natural and human components [47].The landscape scale encompasses a spatial, temporal, and modification dimensions, which are a result of such interactions [47].The spatial dimension includes distinct physical units defined by a national, regional, or local unit.The temporal dimension involves past, present, and future features of the landscape.It also takes into account the ongoing relationship humans have with the landscape, including the effect of their actions on the landscape features and implications for future generations.The modification dimension encompasses the human-driven changes that affected the landscape and its features (e.g., urbanization, deforestation, land rehabilitation).Landscape-scale planning focuses on maintaining the functionality of the landscape by taking into consideration all three dimensions [47,48,81,82].Hence, the main objective of a landscape-scale planning approach is to maintain, restore, and enhance socio-ecological structures and processes that support, and define, the landscape [81]. The three dimensions that characterize a landscape offer a suitable lens from which sustainability issues confronting small islands can be addressed, because they can also account for their dynamic internal and external independences.For example, sustainability-defined as the accomplishment of the broadly accepted sustainable development goals [83] and conservation targets [19]-in small islands depends on embedded trade-offs between maintaining islands' insularity and ensuring socio-economic wellbeing of their residents [1].On the one hand, the sustainability of small islands' socio-ecological systems requires maintaining their insularity, which comprises biological isolation (e.g., avoiding and controlling the introduction of invasive alien species) [2,3] and cultural processes (e.g., cultural identity, coastal/sea lifestyle) [52,84].This necessarily implies certain restrictions and ceilings to the type and intensity of activities affecting their socio-ecological systems, including locals' and visitors' access to certain island features [1,2,52].On the other hand, local socio-economic welfare, a cornerstone in sustainability, requires access to external resources (e.g., technical, financial, goods, services) from mainland and global sources [11,85].However, small islands' interaction with external resources (e.g., foreign ecosystems, cultures, markets) could jeopardize the integrity of their ecological processes and change local culture [1,52,66,86].Baldacchino [51] refers to this dichotomy as the "openness-closeness" or "global-local" small islands dilemma.Furthermore, as intergovernmental organizations, non-governmental organizations, and scholars continue to lean towards the sustainable use of small islands' biodiversity through tourism-compatible conservation, it becomes urgent to build local technical and organizational capacity for managing tourism, socio-economic growth, and conservation according to their own perceptions and aspirations [1,11,14,51,52,87]. Thus, policy integration in the context of small islands needs to take into account these delicate relationships involving internal and external processes, so as to enhance, and not compromise, their socio-ecological systems.In this context, planning at the landscape-scale has enormous potential for policy integration seeking small islands' sustainability, because it is a planning approach that can consider both the relationship between external and internal, and within internal processes affecting their fragile socio-ecological systems. While there is no agreed definition of what policy integration is, or how it can be effectively achieved, scholars usually concur with Underdal's propositions, and identify three main criteria that underpin policy integration: • Comprehensiveness-the extent to which the scope of policy premises match that of policy consequences in terms of spatial and temporal dimensions, and interdependent issues and stakeholders [35,36]; • Aggregation-the extent to which the overall wellbeing of the system as a whole is considered as a premise for the evaluation of policy alternatives and decision making [35]; and • Consistency-the extent to which a policy framework involves all policy actors across sectors and levels of government for the implementation, monitoring, and evaluation of policies to avoid duplications, contradictions, and gaps [35,36,68]. While the three criteria have clear alignment with the concept of landscape-scale planning, to enable an in-depth document analysis and manage the paper length, the analytical framework presented in this paper further explores this potential only for the comprehensiveness criterion.The paper also acknowledges that other analyses focused on the other two criteria are also important and should be the subject of future studies.The comprehensiveness criterion comprises four parameters that are well aligned with the abovementioned concept of landscape-that is, spatial scale, temporal scale, interdependent stakeholders, and interconnected issues (see Figure 1).Firstly, a comprehensive spatial scale for policy integration involves defining the jurisdictional areas policies will apply to, along with the areas of their inherent outcomes and related socio-ecological effects [35,36].The spatial dimension of a landscape contains the hierarchical social and ecological interactions that define it.Hence, it is the basic and functional spatial unit (e.g., a single island).Moreover, due to the landscape spatial scale focusing on functionality, a landscape can be connected with others to form larger ones (e.g., an archipelago) [47,88].Thus, a landscape-scale planning approach aligns with the comprehensive spatial scale element of policy integration, because the landscape scale encompasses the whole area where policy consequences unfold.When policy consequences trespass the spatial jurisdiction where policies are formulated, a landscape-scale planning approach can bring together these jurisdictions to the same governance level; therefore, taking the decision making to its higher level to deal with such policy consequences.This enables the flexibility to allocate decision-making power at the level of governance where it is most effective [47,88], which is largely recommended in policy integration theory [35,36]. Secondly, a comprehensive timeframe for policy integration comprises the consequences of policy measures in the short-to long-term [35].This is particularly important as prolonged consequences in the mid-and long-term affect larger areas, and involve more stakeholders and issues [35,36].Additionally, the definition of an appropriate timeframe for PI implies the right timing between the socio-ecological processes, the generation of scientific knowledge, and socio-ecological matters, that inform policymaking processes [36].Similarly, the landscape temporal dimension encompasses both past socio-ecological interactions and aspirations for the future [47,89].This means that people inhabiting a landscape with an understanding of its past, larger temporal dimension in mind are more likely to project long-term goals/aspirations [89].This attitude could enable the long-term and sustained planning that is needed to accomplish sustainability in socio-ecological systems. Finally, both the comprehensive consideration of interdependent stakeholders and issues for policy integration align with the landscape human-driven modification dimension.Because people shape the landscape through everyday interaction, it is of crucial importance that perceptions and aspirations of landscape users inform decision making [35,36].People's participation is a determining factor of the effectiveness and efficiency with which policies are formulated, implemented, monitored, and evaluated [47,[90][91][92].However, people's interactions with the landscape are underpinned by their evolving, dynamic, and often conflicting value systems [88,89,[93][94][95].Thus, the landscape and its features are never fixed, but constantly evolving [88,95], uncertain, and complex [47,88,96].The comprehensive consideration of interdependent issues involves the combination of perceptions, and local and technical knowledge that can facilitate the identification and addressing of issues [35,36].By acknowledging the dynamic and ever-changing interaction between people and the landscape, the landscape-scale planning approach adopts a flexible and inclusive attitude towards planning that is conducive to comprehensive policy integration, by facilitating collaborative [97] and values-led planning [93,[98][99][100].Thus, it can also modify top-down/centralized decision-making processes [47]. Methodology This study adopts a case study approach [101] focused on the Galápagos Islands to investigate, through document analysis, the extent to which policy integration is facilitated in Galápagos' current institutional arrangements.The selection of documents included in the analysis was guided both by their influence on, and support for, the Special Regime of Galápagos (see Appendix A Table A1).The Special Regime of Galápagos is a unique institutional arrangement within Ecuador (applies only to the Galápagos province) that seeks the conservation of the Galápagos Islands as a World Heritage Site.Thus, documents were interrogated through content analysis [102] to • Understand the context of the case study • Map institutional arrangements, and Explore the structure, scope, goals and policies of the planning instruments. Based on the results, the analytical framework is applied to determine the extent to which Galápagos institutional arrangements, expressed through the legal and planning frameworks, meet the four parameters of the comprehensiveness criterion of policy integration (spatial scale, timeframe, interdependent stakeholders and issues).Then, the study discusses the feasibility of considering the spatial, temporal, and human-driven modification dimensions of a landscape-scale planning approach to advance the parameters of the comprehensiveness criterion of criterion of policy integration.Subsequently, recommendations to trigger the implementation of a landscape-scale planning approach in Galápagos are suggested, along with relevance for other small island systems. Case Study: The Galápagos Province Galápagos is one of the 24 Ecuadorian provinces, and is located approximately 1000 km to the west of the Ecuadorian mainland on the Pacific Ocean.The territory of the Galápagos province includes protected areas and human settlements (see Figure 2).The protected areas comprise the territories of the Galápagos National Park and the Galápagos Marine Reserve.The Galápagos National Park covers 97% of the combined land surface of all the islands, islets, and rocks that form the Galápagos Archipelago, except for the human settlement areas.The Galápagos Marine Reserve covers the totality of internal waters between islands and a 60 nautical mile buffer around the baseline of the islands.The human settlements comprise three urban and five rural areas in the islands of San Cristobal, Santa Cruz, Isabela, and Floreana which cover the remaining 3% of Galápagos' land surface, and host an estimated population of approximately 30,000 people [103].Galápagos Archipelago is worldwide known for its unique and highly endemic wildlife, leading to the designation of Galápagos National Park as a World Heritage Site in 1978 [105].Later in 2001, the Galápagos Marine Reserve was included in the World Heritage Site designation [105].Additionally, Galápagos was declared a Biosphere Reserve by the UNESCO in 1984, and the south wetlands of Isabela Island were recognized as a RAMSAR Site in 2002 [106].These designations compel the conservation of the Galápagos' unique ecosystem by targeting sustainability in all anthropogenic activities that take place in the islands.However, the Galápagos' socio-ecological system is currently under great and increasing pressure from tourism, fishing, unplanned development, and a raft of interconnected social, economic, and environmental problems derived from these activities (see Table 1).The Special Regime of Galápagos was therefore enacted in 1998 to ensure the protection of Galápagos' socio-ecological system and associated wellbeing of Galápagos' residents [102]. Tourism Galápagos is experiencing an accelerating growth in the number of visitors since the late 1960s [7,103].The number of tourists arriving to the islands every year went from a couple of thousands in 1969 [7] to over 220,000 in 2015 [103]. 1. Encourages the migration of Ecuadorians and foreigners who seek to profit from tourism and its associated economic activities [7,62,103,107].2. 3. Increases the demand for goods and services (e.g., food, fuel, commodities) from the mainland to support tourism-related activities [103].4. Increases the risk of introduction of invasive alien species due to more cargo ships and planes arriving to the islands [103]. 5. Intensifies the intersectoral competition for greater access to the Galápagos National Park and Galápagos Marine Reserve causing claims for a more equitable distribution of tourism revenue across local communities [108,109].6. Encourages associations and companies of cruise ships, land-based accommodation and food services, cabotage, terrestrial transport, and ecotourism guidance in every island to compete for visitors, which has a direct impact on the urban development of the islands [64,103]. Fishing Since the 1980s, fisheries quickly intensified driven by the demand of international markets for lobster and sea cucumber, particularly from Asia [7,29,64]. 2. Encourages the illegal fishing of sharks within the Galápagos Marine Reserve by residents and large foreign vessels alike [7,29,64]. 3. Triggers opposition to conservation and tourism amid claims of displacement, inequity, lack of livelihood alternatives, and insufficient technical and financial support by environmental agencies [7,54,64].4. In several occasions, fishermen's discontentment with conservation measures resulted in strikes and violence.Such measures managed to directly influence the sacking and appointing of Galápagos National Park Service' directives [105]. 5. Contributes to seasonal immigration and population growth [7], and as in the tourism sector, different fishing associations were formed in every island [103]. Conservation/Governance Galápagos presents a historical lack of institutional/organizational/technical capacity when implementing and enforcing policy frameworks and institutional arrangements [7,54].Also, there is great instability in authority positions [105]. 1. Causes struggles between government agencies and the fishing and tourism sectors, damaging relationships and trust [56,105,111]. 2. Results in poor public participation/engagement in policymaking (cantonal assemblies: non-governmental structures for public participation remain inactive) [103]. Human settlements Institutional arrangements [112,113] limit municipalities and rural parishes resources and capacity, compared with provincial governments and national ministries, but are left to deal with the consequences of population growth. 1. The growing number of residents and visitors accentuate issues within the human settlements, such as the deficit of freshwater in most of Galápagos inhabited islands [59], and the pressing need for appropriate systems to treat wastewater [57] and dispose of solid waste [58]. 3. Land use and development plans are not implemented or followed. Institutional Arrangements for Galápagos Understanding Galápagos' institutional arrangements requires an in-depth analysis of the extensive and intricate hierarchy of laws and planning instruments governing the archipelago.Firstly, Article 242 of the 2008 Ecuadorian Constitution [113] establishes that the State's territory is subdivided into regions (a recently created level of governance), provinces, cantons (municipalities), and parishes.Secondly, the hierarchy and articulation of all the planning instruments from national to parish levels is set by the National Secretariat of Planning and Development (SENPLADES) [114,115] (see Figure 3b).Additionally, the Organic Code for Territorial Organization, Autonomy and Development (COOTAD), and the Organic Code for Planning and Public Finances (COPFP) establish that in order to access Central Government funding, all GADs must formulate plans for the development and territorial organization under the methodology and planning template of the SENPLADES [112,114,116].Furthermore, Article 280 of the Constitution demands all GADs' planning instruments to be aligned with the objectives and policies of the National Plan for the Good Living (PNBV) [113].To assist in the implementation and adaptation of the National Plan for the Good Living to lower territorial levels there are national ministerial and regional planning instruments.These include the zonal agendas, which are part of the National Territorial Strategy (included in the PNBV), and have provisions for the development and territorial organization (land use/spatial planning) of adjacent provinces that are functionally linked through social, economic, and ecological processes [117,118].Additionally, as stipulated in the Constitution and the COOTAD, zonal agendas intend to assist provinces with their responsibility to form regional GADs [112,113].Under the current provisions, Galápagos belongs to Zone Five, together with the provinces of Guayas (except for the cantons of Guayaquil, Samborondon, and Durán), Los Ríos, Santa Elena, and Bolivar [117].Nevertheless, as shown in Figure 3a,b, no regional GADs have been formed thus far.Hence, there is no government agency currently in charge of the formulation of regional norms or the implementation of Zonal Agenda 5. The Galápagos Special Regime (GSR) The GSR was first established under the Article 239 of the Ecuadorian Constitution of 1998 [119], and later ratified under Articles 242 and 258 of the Ecuadorian Constitution of 2008 [113] to ensure the conservation and sustainable management of Galápagos.Additionally, the 2008 Constitution stipulates that the Special Regime of Galápagos is to be governed under the Organic Law for the Special Regime of the Province of Galápagos (LOREG), and in alignment with the national and subnational legislation and planning instruments [113].Under Articles 4-8, the LOREG establishes that the Government Council for the Special Regime of Galápagos (CGREG) and its Technical Secretariat are the institutions in charge of planning and policy implementation at a provincial level.However, decision making takes place at the meetings of the CGREG Plenary Board. Under Article 10 of the LOREG, the CGREG Plenary Board consists of The President of Ecuador, who is represented by an appointed minister as head of the CGREG (who also chairs the Plenary Board); the ministries of Environment-represented by the Galápagos National Park Service, Tourism, Agriculture, and the SENPLADES; the municipal GADs of San Cristobal, Santa Cruz, Isabela; and one representative from the five Galápagos' perishes GADs (see Figure 4).While only CGREG Plenary Board members have decision-making power, there are other institutions that provide technical assistance and advice.Some of the most important ones includes the Ministries of Transport and Infrastructure, Health, and Education, the Ecuadorian Navy, the Charles Darwin Foundation, and NGOs, such as Conservation International and Galápagos Conservancy.Additionally, any individual or group can participate in the meetings and debates of the CGREG Plenary Board and GADs (without voting power), requesting the empty chair mechanism (a public participation mechanism established under Article 101 of the Constitution). Furthermore, the Regulations for the Implementation of the LOREG (draft) [120] and the Galápagos' Territorial Organization and Sustainable Development Plan [103], respectively establish two means for public participation: the Consultative Management Board for the stakeholders involved with the Galápagos Marine Reserve, and the Cantonal Assemblies for stakeholders involved with urban and rural areas.However, these are not operational to date.Other important aspects of the LOREG 2015 include the provisions of (i) Articles 33 and 34, that subject all the government agencies and GADs' planning instruments to the CGREG's Sustainable Development Plan; (ii) Articles 35-49, that establish four residential status (permanent resident, temporary resident, tourist, and transient) and the criteria for their respective eligibility, rights, and restrictions; (iii) Articles 50-78, that set the parameters for regulating socio-economic activities (e.g., tourism, fishing, agriculture, and craftsmanship) in the islands, which convey preferential rights to permanent residents; and, (iv) Articles 85-113, that define the jurisdictions for the institutions in charge of the governance and management of the Galápagos Marine Reserve, Galápagos National Park, Galápagos' province, Galápagos' urban and rural areas, and Ecuadorian Territorial Sea. Under articles in section IV, the management of the Galápagos National Park and Galápagos Marine Reserve corresponds to the Ministry of Environment and its branches, including the Galápagos National Park Service and the Galápagos' Biosecurity Agency under environmental laws.Similarly, the regulation of fishing, tourism, agriculture, freight, and interisland cabotage fall over the corresponding ministries and secretariats, according to organic and ordinary laws.Additionally, the jurisdictions/responsibilities over the different territorial levels is aligned with the provisions established in the Constitution 2008 (see Appendix B, Table A2).Nevertheless, the regulation and control of immigration are the responsibility of the CGREG and its Technical Secretarial.Furthermore, the CGREG Plenary Board-informed by the corresponding ministry, secretariat, or GAD-is the only entity allowed to grant sea or land-based licenses for tourism.Finally, while each agency oversees the application of administrative sanctions according to provisions of the LOREG and other environmental and civil laws, enforcing these laws and ensuring Galápagos' safety is the responsibility of the Ecuadorian Navy for the Galápagos Marine Reserve and Ecuadorian Territorial Sea, and of the Police and the Ecuadorian Army for land areas. Assessment of Comprehensiveness of Policy Integration Galápagos' institutional arrangements show evidence of endeavors of a landscape-scale approach that could facilitate some degree of policy integration.These include the notions behind the formation of regions (even when these remain unformed), and more importantly, the establishment of the CGREG and the CGREG Plenary Board.On the one hand, the CGREG Plenary Board has the potential to become a suitable arena for parishes, cantons, provincial and national stakeholders representing multiple sectors to assess common issues, debate alternatives, define goals, and formulate policies and plans to achieve them.On the other hand, the CGREG has been conveyed the authority to implement and enforce the CGREG Plenary Board's decisions.As result of these arrangements, important decisions have been made, and measures implemented to increase immigration and biosecurity controls.Additionally, the CGREG Plenary Board has distributed the funds from the tourists' entrance fee to Galápagos in the following manner: 50% for the Galápagos National Park Service and Galápagos' Biosecurity Agency, 25% for the three municipal GADs, 20% for the CGREG, and 5% for the five parishes GADs [122].Lastly, both legal and policy frameworks support and encourage decision making and public engagement at all territorial levels (at least in a rhetorical manner).Nonetheless, there are some significant shortfalls that need to be addressed if effective policy integration for the sustainability of Galápagos is to be achieved, as discussed next. Comprehensive Spatial Scale The jurisdictions of the territories, and their respective institutional arrangements for decision making, follow political divisions that do not represent comprehensive functional socio-ecological structures and processes.For example, Galápagos depends on the Ecuadorian mainland for the provisioning of goods and entry of tourists through the routes that connect the maritime ports and airports in the archipelago with Quito (in Pichincha Province) and Guayaquil (in Guayas Province).These are the only entry points of cargo and people to the islands [103,123].However, being Metropolitan Districts, Guayaquil and Quito constitute administrative zones in their own, Zone 8 and 9, respectively [117].Hence, in addition to the absent regional GAD to administrate Zone 5, spatial comprehensiveness is constrained because administrative ties, including official and regular collaboration, need to be established with Quito and Guayaquil.To magnify this problem, due to no regions having been officially formed to date, Article 4 of the Organic Law to reform the COOTAD states that provinces which have not been organized into regions by 2028 will be incorporated into regions under the initiative of the Presidency and the approval of the National Assembly [124].Moreover, in case a Central Government-driven regionalization supersedes the current zonal division of the National Territorial Strategy, there could be an institutional mismatch that further exacerbates the current governance/administrative inconsistencies.This gap could have severe impacts in the governance and management of the islands, and in the integration of Galápagos' aspirations with the rest of Ecuador. Additionally, the Galápagos Marine Reserve with the GADs' coastal areas are linked by a broad range of socio-ecological processes (e.g., transportation, tourism, fishing, recreation, sewage treatment, species migration, and hunting patterns) [103,123,125].Nevertheless, these are scarcely addressed by the CGREG and GADs' planning instruments.Programs to protect iconic species and places (e.g., sea lions, mangroves, beaches, jetties) do little to cover the full extent of the land-sea permeability that occurs through the GADs' coastal areas [123].On the one hand, GADs' measures for construction, transportation, noise regulation, and even street lighting, could have an impact on nearby marine areas.On the other hand, Galápagos National Park Service, Ministry of Tourism, and Navy's measures to regulate maritime activities could have a socio-economic and cultural impact on urban areas.However, provisions to respond to, or alleviate, such impacts are missing from current planning instruments.The only provisions for an integrated land-sea management are included in the Galápagos National Park Service's Management Plan, but these are limited to the protected areas of the Galápagos National Park and the Galápagos Marine Reserve.Moreover, there are no permanent mechanisms for establishing the collaboration between the GADs and the Galápagos National Park Service outside the CGREG Plenary Board.Instead, their collaboration takes place through specific programs and projects whose budget and monitoring are not always specified in the planning instruments.Additionally, the Navy, which is a key stakeholder in the regulation and control of maritime activities, is not a member of the CGREG Plenary Board. More importantly, there are also socio-economic processes (e.g., transportation, commerce, fishing, tourism, political) linking the urban and rural areas of the four inhabited islands that are not considered in the GADs' plans.On the contrary, each municipal GAD presents its own approach to urban development.Moreover, GADs seem to be engaged in an open competition to become the most popular destination for tourists within Galápagos.The same competition is also reflected in the attitudes of fishing associations and companies, tourism operators, artisans, and transport operators that exist in each island.Subsequently, these island-constrained objectives make the development of a common front for addressing cross-cutting issues affecting Galápagos highly unlikely.Additionally, they may hinder the implementation of the CGREG Plenary Board policies and programs. Comprehensive Temporal Scale From a temporal scale perspective, one of the main barriers for effective policy integration in Galápagos comes from the mismatch between the short-term lifespan of the planning instruments and their unrealistically ambitious long-term goals.Following the SENPLADES planning methodology, the CGREG and GADs' planning instruments try to adjust to the four-year period of democratically elected authorities [114,115].This relatively short-term planning period, together with a potential new government with different agendas every four years, obstruct the monitoring of policy measures over time, and jeopardize the accomplishment of long-term goals, such as sustainability, equity, and social justice.Almost all planning instruments relevant to Galápagos aim to achieve sustainable development on their own right over a four-year timeframe.This is a recurrent trend among ministerial, provincial, cantonal, and even parishes' plans.There is also a lack of scaled down goals and targets (e.g., achieve 60% of occupation in accommodation services or increase the share of tourism-related profit that stays in the Galápagos territory), with goals broadly focused on achieving sustainable ecotourism. Only the Santa Cruz GAD plan surpasses a four-year timeframe.Additionally, except for the Galápagos National Park Service Management Plan, the CGREG, and GADs' planning instruments present little to no evidence of the assessment of policy consequences, or of provisions for their articulation in subsequent plans. There is also bad timing between legal and policy frameworks.Laws and ministerial plans (including ministers) are often changed under the directives of new Presidents and National Assemblies that renew every four years.However, there is a two-year lag between the election of national authorities and GAD authorities.This lag is translated into the planning instruments that remain aligned to previous plans for half of their lifespan.In Galápagos' case, the lifespan of most of the GADs' planning instruments in this analysis is from 2012 to 2016.Thus, these were formulated before the enactment of the LOREG 2015, and the formulation of the CGREG Sustainable Development Plan (2015Plan ( -2020)).A content analysis of the latter revealed that its objectives and policies were not informed by the planning instruments at lower territorial levels.Alternatively, the CGREG plan implements a top-down approach that should be reflected in the GADs' plans, relieving the current ones for the term 2016-2019.Nevertheless, this information is not available to date on the GADs' websites or through the CGREG's reports.Moreover, the Galápagos National Park Service Management Plan (2014Plan ( -2017) ) and the Santa Cruz GAD plan (2012-2027), to date, have not been reviewed after the LOREG and CGREG plan were established. Finally, the planning cycle is happening at a faster rate than the generation of technical and scientific information/knowledge.The Galápagos' short-term plans are implementing some drastic measures attempting to solve immediate problems regarding infrastructure and the provisioning of services.For instance, Santa Cruz GAD and CGREG's plans present provisions to consolidate urban areas through incentives to develop empty lots [103,126].In Santa Cruz, this has resulted in new development areas in El Mirador sector that nearly doubled the area of the existing settlement [103].Also, all the territorial organization plans present provisions to create more tourism-related infrastructure, which is clearly polarized around the coastline of urban areas [103,[126][127][128].These will have a long-lasting impact on Galápagos' socio-ecological system that will not have yet unfolded by the time new policies are implemented. Comprehensive Stakeholder Interdependence There are two main factors influencing public engagement in Galápagos: (i) previous troubled relationships between government agencies and local fishing and tourism sectors; and (ii) legacy of past struggles that resulted in inefficient governance and planning structures.In 1998, when the first LOREG was enacted, it contained provisions for public engagement in policymaking.A key attribute included in this law was the creation of two platforms for the participative management of the Galápagos Marine Reserve.Firstly, the Participative Management Board (JMP-Spanish acronym) in charge of management decisions at provincial level was integrated by the Galápagos National Park Service, Navy, fishing and tourism guides associations, the Charles Darwin Foundation, and other representatives of ministries and social groups.Secondly, the Authority of Inter-institutional Management (AIM-Spanish acronym) was in charge of ratifying and enforcing the decisions of the JMP from a ministerial level [53,129].Although these platforms achieved relative success in terms of policy actors' participation [53], the enforcement of the decisions was less effective, due to the limited capacity of government agencies [54,105].The unsolved conflicts between conservationist and fishing sectors hindered agreements and the implementation of management plans for the Galápagos Marine Reserve and Galápagos National Park [7,52,64,66] (see Appendix C, Table A3 for a chronology of anthropogenic activities in Galápagos).Ultimately, these factors were determinants in the inclusion of Galápagos in the list of Endangered World Heritage Sites from 2007 to 2010 [105].Consequently, government agencies were restructured and entrusted with more decision-making power [130], and the recently enacted LOREG 2015 set new institutional arrangements for the governance and management of Galápagos.Under these measures, Galápagos' institutional arrangements present a clear top-down approach that has restricted the direct participation of key stakeholders.While there are approximately 70 stakeholders comprising government, civil groups, and NGOs [103], only nine of these have decision-making power through the CGREG Plenary Board (five are appointed ministers, and four are democratically elected representatives of the urban and rural areas).Although provisions to grant citizens the right to participate in policymaking and planning at all levels of government are specified in the Constitution (Articles 95-101) [119], there are no clearly defined mechanisms for its implementation.For instance, the CMB and the cantonal assemblies are not operational.Moreover, even when these become operational their resolutions will not be legally binding. To aggravate the situation, not all stakeholders can afford the costs involved in participating in the CGREG Plenary Board.Additionally, the absence of an open platform to enable the communication between civil groups with government agencies in each island, and between civil groups across the islands, has contributed to the development of island-specific planning approaches.Another important fact is that planning instruments do not show evidence of public participation.The only two exceptions found in this analysis are the Bio-Agriculture Plan of Ministry of Agriculture [131] and Isabela GAD Strategic Tourism Plan [132].These plans documented public participation and have objectives and policies that are more consistent with the local socio-economic and socio-ecological contexts.Notably, these plans do not follow the SENPLADES planning methodology. Comprehensive Issues Interdependence The way in which spatial scales, policy timeframes, and stakeholders are considered plays a determinant role in the amount of interdependent issues that can be addressed under Galápagos institutional arrangements.Firstly, the Zonal Agendas-related administrative constraints represent a major obstacle to address pressing issues affecting the islands.These include the shortage of food, the lack of infrastructure to implement quarantine measures, and the need to maintain and improve both the regulation and monitoring of tourists and residents entering and exiting the islands [103].Secondly, the policies and programs in the GADs' planning instruments respond to pressing local issues, mainly related to the provisioning of basic services (e.g., drinking water, sanitation, waste management, urban development and housing).Conversely, provincial plans are focused on improving governance and institutional strengthening.This means that core sustainability-related issues, such as climate change adaptation, education, capacity building, and public health, are not considered by planning instruments.The limited range of issues considered in these instruments leads to a limited number of stakeholders being engaged in the process.There is also no common front to address the common goals for inter-island connectivity and cross-cutting issues, because planning instruments are predominantly locally oriented and lack long-term, strategic focus. Finally, Galápagos' public institutions have struggled with adapting the few sustainable goals set at national level to the local socio-ecological context of the archipelago.Thus, the same broadly defined national-level goals are repeated throughout provincial, cantonal, and parishes plans.This is because fundamental sustainability-related concepts such as conservation, ecosystem services, eco-tourism, public participation, and island culture may not be clearly understood by local and even national policy actors.For instance, the narratives of most planning instruments use the term conservation as a synonymous for preservation.In this sense, the conservation of Galápagos is seen as the non-intervention of nature.Under this premise, the only acceptable economic alternative left is a utopian view of eco-tourism that only brings economic benefits to the locals and has no side effects.Similarly, the understanding of public participation is limited to informing stakeholders of decisions.This hinders debates, consultation, and even monitoring of how effective these policies might be.Additionally, the few provisions for culture only consider artistic expressions diminishing the role of identity or sense of place.One of the main problems with these poorly understood concepts is their failure to recognize change.Acknowledging and preparing for the changes embedded in socio-ecological processes is crucial for islands' sustainability, especially because islands are affected by global trends in climate and market demands over which they have no control [1,14].Hence, without multidisciplinary teams that facilitate the understanding and the incorporation of such concepts into planning and policy implementation, it is unlikely that the long-term sustainable goals of the PNBV and ministerial plans will result in policies, programs, and projects that effectively contribute towards Galápagos' sustainability. Discussion Achieving sustainable development goals require orchestrated efforts from multiple sectors, institutions, and stakeholders [14,85,111].In the case of small islands, such as Galápagos, sustainability can only be achieved by comprehensive policy integration efforts that seek to manage integrated land and sea issues, taking a socio-ecological system perspective that brings together government levels and society.This requires a unifying platform that accommodates spatial and temporal scales, interconnected stakeholders, and interdependent issues in an articulated and functional manner.A landscape-scale planning approach could provide such unifying platform because the landscape is a dynamic unit with physical (spatial and temporal) dimensions and changing socio-ecological interactions [48]. Applying a landscape-scale planning approach to facilitate policy integration for the conservation of Galápagos implies a careful consideration of the elements underpinning its socio-ecological processes and how these unfold across time and space.It is also important that future decision making is informed by a better understanding of how policy measures affect these processes, as this is critical for enhancing its sustainability [36,37]. In regard to the spatial dimension, this study found that Galápagos institutional arrangements are driving territorial compartmentalization rather than much necessary integration to solve the raft of complex internal and external issues affecting its conservation.Under a landscape-scale planning approach, such fragmentation in the Islands' governance and management could be overcome with the creation of critical partnerships entrusted with official and permanent channels of communication.For example, a partnership between the CGREG and the Metropolitan Districts of Quito and Guayaquil could be created to ensure locally relevant issues affecting the Islands are not overshadowed by other pressing issues affecting these metropolitan areas.A partnership between the Galápagos National Park Service, the Navy and the GADs could improve the management of land-sea areas, especially considering that both the Islands' tourism attractions and socio-ecological systems are highly dependent on a healthy marine environment.A partnership between stakeholders in each and across islands could improve collaboration and equal distribution of opportunities and income across different sectors and social groups, as opposed to the current competition that is leading to the depletion of the Islands' natural resources and fragile ecosystems.These partnerships need to be supported by well-funded programs to improve Galápagos' institutional and social capacity, and enable inclusive stakeholder participation.An existing example of partnership of this nature is the long-term agreement between the Charles Darwin Foundation (international NGO) and the Galápagos National Park Service to undertake scientific research and provide technical assistance that informs decision making [123].Furthermore, the multidisciplinary research carried out by the Charles Darwin Foundation could be extrapolated to set guidelines for multidisciplinary work between government institutions and with non-governmental groups/associations.Ultimately, this has the potential to lay the foundation for future interdisciplinary approaches. Additionally, by considering the temporal dimension, a landscape-scale planning approach would ensure Galápagos' current and future planning instruments are focused on long-term goals that guide objectives and programs adjusted to their short-term lifespan.Furthermore, these plans would also incorporate provisions for ongoing follow-up, monitoring, and update to deal with the ever-changing character of landscapes, and incorporation of best available knowledge to inform the Islands' management of their fragile socio-ecological systems.Finally, the issues to be addressed through these plans would be informed by a raft of stakeholders and the partnerships formed by these.The collaborative planning character of a landscape-scale planning approach could also facilitate the creation of arenas for deliberation to overcome past struggles and conflicts.However, effective implementation of all these measures entails changes in the socio-political structure and culture.In particular, stakeholders from all territorial levels and across all sectors must be aware that there are pros and cons embedded in conservation, tourism, fishing, and development.More importantly, they need to play a greater role in making collective and well-informed decisions to enhance the sustainability of the Islands' socio-ecological systems. Lastly, through the consideration of the human-driven modification dimension, the collaborative planning character of a landscape-scale planning approach could also be instrumental in addressing one of the most pressing needs in the Galápagos Islands, namely building local capacity (both institutional and social) to better manage its socio-ecological systems [54,103,123].Currently, skills shortage in the Islands is being addressed by hiring professionals from the mainland enabled through regulations set by the LOREG which grant them temporary residency status.In the best-case scenario, these professionals are likely to contribute to form multidisciplinary teams to implement the above-listed opportunities both effectively and efficiently.Nevertheless, they are likely to compete with permanent residents for jobs, bring in a different set of values relating to lifestyle and culture, and increase the demand for already scarce resources (e.g., water, food, housing) [103].Their temporary residency status only allows them to work under yearly renewed contracts that determine the type of work they can do.Furthermore, temporary residents and their family members are not allowed to buy land in the Islands.These restrictions narrow the range of professionals that could be attracted to work in Galápagos, diminishing their longer-term commitment with the Islands' issues, and creating a social divide between temporary and permanent residents.A collaborative landscape-scale planning approach could ensure temporary skilled migration programs had long-lasting capacity building benefits for Galápagos by facilitating the development of a nourishing interaction between temporary and permanent residents.This does not preclude the need for educational and training programs specially targeted at permanent residents to prepare them for highly skilled jobs in the mid-to long-term future.Additionally, it would create opportunities for both groups to deliberate about, and better understand, the impacts of such programs on Galapagan culture, identity, and lifestyle (insularity), so as to not compromise these. The same reasoning can be applied to the benefits and problems associated with activities relating to the tourism, urban development, and conservation sectors, hence enabling stakeholders to define the boundaries of their insularity and make the necessary decisions to balance it with the achievement of their well-being based on a sound socio-ecological system, which is the main premise of sustainability, and must be the first and foremost objective of policy integration.Nevertheless, measures to achieve this objective are missing from current Galápagos institutional arrangements and planning instruments.By focusing on the everyday interactions that occur between stakeholders in a landscape, it is possible to understand people's perspectives, concerns, aspirations, and value systems, and use these as valid inputs to inform decision making.This is perhaps the main contribution a landscape-scale planning approach could bring to the sustainability of Galápagos and other islands facing similar issues. In this context, there is potential/need for adopting a landscape-scale planning approach because it may facilitate the interconnected processes of (i) building Galápagos' social and institutional capacity to deal with ongoing social and environmental change; (ii) recognizing, restoring, maintaining, and enhancing Galápagos' socio-ecological processes; and, (iii) facilitating the development of sustainable communities in the Archipelago. Under the current institutional arrangements, Galápagos presents a top-down approach to planning, policymaking and governance.In order to achieve the effective implementation of a landscape-scale planning approach this study suggests that the Consultative Management Board and cantonal assemblies become operational, and assign funds to facilitate the participation of stakeholders in the CGREG Plenary Board meetings.In addition, it is suggested allocating funding for permanent platforms dedicated to facilitate the communication between non-government stakeholders across islands.Finally, it suggests the development of programs focused on amending bad historical relationships among stakeholders and training/educating stakeholders on landscape planning topics.These include, for example, the concept of landscape, implications of landscape-scale planning, sustainability, ecosystem services, and resilience, to name a few.Furthermore, particular attention should be paid to improving the understanding of the levels and implications of public participation in planning [133].These small but necessary steps could be incorporated in Galápagos planning instruments as early as possible, and become part of planning platforms of future sectional elections for cantons (equivalent to municipalities) and parishes.Subsequently, these steps could trigger the implementation of a landscape-scale planning approach that improves the consideration of the four parameters of the comprehensiveness criterion of policy integration.Ultimately, more effective policy integration accompanied with progressive and sustained local capacity building could facilitate the accomplishment of long-term sustainable development goals. Lastly, the recommendations suggested in this article are the result of an in-depth context-based analysis of the Galápagos Islands case study.Transferring these findings and applying the methods of this study to other small island or mainland contexts would necessarily require a thorough evaluation of their respective legal and planning frameworks and socio-ecological context.In each case, all these local factors will determine the most effective path to implement a landscape planning approach to improve policy integration.Furthermore, the premises to achieve sustainability might differ from place to place. Conclusions The Galápagos Islands' case study outlined three main themes regarding the sustainability of small islands.Firstly, small islands are functionally linked to larger mainland systems.Therefore, planning, policymaking, and governance arrangements must necessarily take these links into consideration.Secondly, in order to ensure that the perceptions and aspirations of small islands' inhabitants inform decision making, there is a pressing need for increased local technical and organizational capacity.This entails enabling small islands' communities to be aware and prepared for the changes embedded in any given path chosen to ensure the conservation of their insularity and to secure their welfare.Finally, this study illustrated that a landscape-scale planning approach could potentially improve the consideration of the policy integration criteria.However, the insights presented here are based on findings from the document analysis covering legal and planning instruments, and limited to the parameters of comprehensiveness criterion of policy integration.Future studies on the contributions of a landscape-scale planning approach to the remaining policy integration criteria (aggregation and consistency) could improve the understanding of its full potential or feasibility for addressing small islands sustainability-related issues.Moreover, further studies could focus on knowing more about the perspectives and aspirations of non-government and community-based organizations whose livelihood depend on sound ecosystem services, including, but not limited to, fishing groups, tourism services operators, tourist guides, and cultural groups in small islands. Appendix C Table A3.Chronology of anthropogenic activities in The Galápagos Islands based on the stages defined in [7]. Stage Year Main Events Pre-Hispanic Prior to 1492 Galápagos were not inhabited by humans when discovered by the Spaniards.However, there is archeological evidence that indigenous tribes visited Galápagos before.Nevertheless, the origin of the indigenous artifacts found in the islands has not been determined [144]. Extractive Exploitation 1535 Galápagos Islands are discovered by accident when Panama's Bishop was travelling from Panama to Lima [145]. Following Centuries Whalers and buccaneers introduced alien species and depleted native species, particularly the giant tortoises [146]. 1832 The Ecuadorian Government claimed the archipelago to be part of its territory [106]. Colonization 1832-1837 The first colonization tried to establish a penal colony [144]. 1835 British Scientist Charles Darwin visits the Galápagos Islands and studies the very specialized physiological adaptations of finches and giant tortoises across the islands [7]. 1859 Darwin publishes his findings on the book "The Origin of Species by Means of Natural Selection" proving the evolutionary theory [7]. 1869-1878 The second colonization tried to start agricultural activities and commerce in Floreana Island [144]. 1879-1904 A third colonization attempt takes place in San Cristobal Island where Manuel J. Cobos tried to stablish a sugar mill.After this, colonization in San Cristobal, Santa Cruz, Floreana and Isabela continues steadily [144]. During the first half of the 20th Century, the main economic activity was agriculture [64]. 1936 The Archipelago of Galápagos is declared as Protected Area amid requests from the international scientific community to protect their scientific value.Nevertheless, there were no official institutions assigned to the implementation of such measure [64,106]. 1942-1947 The United States establish a military base in Baltra Islands to protect the Panama Channel [144]. 1947-1958 Naturalist Lack and Eibl-Eibesfeldt encouraged the study the ecological system of Galápagos and demanded more effective protection of the islands [147]. 1950s Fishing becomes the main economic activity in Galápagos attracting a new wave of colonizers and shifting resident from agriculture to the seas [64]. 1959 Creation of the Charles Darwin Research Station in Belgium, and the Charles Darwin Foundation and the Galápagos National Park in Ecuador.However, the Boundaries of the Galápagos National Park were not defined [147] Table A3.Cont. Wilderness Conservation 1960 The Galápagos National Park Service and Charles Darwin Foundation started a joint work to protect native species on the islands [64]. 1973 Creation of the Galápagos Province; the boundaries of the Galápagos National Park were fully defined; the first body of rangers became operative; the first plan for the management of the terrestrial area was created [64]. 1974 The first management plan for the Galápagos National Park was formulated and included provisions to protect a 12 nautical mile buffer around the coastline of the islands [64]. 1978 Galápagos Islands were declared as one of the original eight a World Heritage Sites [105]. 1980s Tourism displaced fishing and agriculture and assumed the role of economic drivers in Galápagos [7,62,64].Conflicts between the fishing sector and the tourism and conservation sectors begin [64]. 1984 Woods Hole Oceanographic Institute wrote a report advocating for protection of marine areas within a 15 nautical miles buffer around the islands [7].UNESCO declares Galápagos as Biosphere Reserve [106]. 1986 The report of the Woods Hole Oceanographic Institute resulted in the creation of the Galápagos Marine Resource Reserve (GMRR), a 15-nautical miles buffer around the islands.Nevertheless, the GMRR did not have a protected area status [64]. 1990s During this decade, struggles between fishermen and the Galápagos National Park Service resulted sever conflicts and even violent actions [64,105]. 1994 The Ecuadorian Government unsuccessfully request the addition of the marine reserve to the World Heritage Site [105]. 1998 The First Organic Law for the Special Regime of the Galápagos Province (LOREG) is passed; the Galápagos Marine Reserve, a 40 nautical miles wide buffer around the baseline of the islands, is created; The Galápagos National Institute (INGALA) is created to assume the planning and management of the islands [129]. Conservation-Development 1999 First management plan for the Galápagos Marine Reserve was approved [7]. 2001 The Galápagos Marine Reserve is included as a World Heritage Site [105]. 2002 Isabela's south wetlands are declared RAMSAR Site [106].2003 First regional (provincial) plan for the Galápagos National Park was introduced [7]. 2000s Because of political instability, lack of institutional capacity and coordination the LOREG (1998) was not fully enforced and plans were not implemented.Consequently, conflicts between the Galápagos National Park Service and the fishing sector continued.Also, Galápagos experienced an unprecedented tourism growth causing population growth and increasing the pressure on natural resources [7,52,64,66]. 2007 Galápagos Islands are placed on the Endangered World Heritage Sites List due to failures to control the introduction of invasive alien species, and to regulate immigration and the increasing tourism activities [105]. 2008 The new Ecuadorian Constitution creates the Galápagos Government Council (GGC) to replace the INGALA as the planning and management authority in the Galápagos Province [113]. 2007-2010 Several measures are implemented to address the requirements of the World Heritage Committee and Galápagos Islands are removed from the Endangered World Heritage Sites List [105]. 2014 The first plan for the joint management of the Galápagos National Park and Galápagos Marine Reserve is introduced [123]. 2015 A new Organic Law for the Special Regime of the Galápagos Province is passed [121].The LOREG (2015) triggers the elaboration of new regulations and plans for the conservation of the Galápagos Islands. Figure 1 . Figure 1.Alignment between policy integration comprehensiveness criterion and landscape-scale attributes. Figure 3 . Figure 3. (a) Legal framework for the province of Galápagos; (b) policy framework for the province of Galápagos.Laws and planning instruments are hierarchically arranged from top to bottom and from left to right.Continuous lines show direct influence over other legislation while dashed lines show indirect influence.Acronyms: PNBV-National Plan for the Good Living; LOREG-Organic Law for the Special Regime of Galápagos; CGREG-Government Council for the Special Regime of Galápagos; GAD-Decentralized Autonomous Government; GNPS-Galápagos National Park Service. Figure 4 . Figure 4. Institutional arrangements and stakeholder interactions in the Province of Galápagos [103,113,120,121].Thicker lines show strong interactions, while thinner lines show weak interactions.Continuous lines show permanent/established interactions, while dashed lines show intermittent interactions or inoperative links.Red lines show public participation access in CGREG or GADs meetings through the empty chair mechanism.Acronyms: WHC-World Heritage Centre; CBD-Convention on Biological Diversity; CGREG PB-Government Council for the Special Regime of Galápagos Plenary Board; GNPS-Galápagos National Park Service; CDF-Charles Darwin Foundation; CMB-Consultative Management Board; GAD-Decentralized Autonomous Government.	2018-12-12T11:50:28.423Z	2018-04-17T00:00:00.000	{ "year": 2018, "sha1": "0c74a20a8a650a196f8ba01aa851084614702968", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2071-1050/10/4/1228/pdf?version=1525348901", "oa_status": "GOLD", "pdf_src": "ScienceParseMerged", "pdf_hash": "0c74a20a8a650a196f8ba01aa851084614702968", "s2fieldsofstudy": [ "Environmental Science", "Political Science", "Geography" ], "extfieldsofstudy": [ "Economics" ] }
257779076	pes2o/s2orc	v3-fos-license	Oxidants and Cardiorenal Vascular Remodeling—Insights from Rare Genetic Tubulopathies: Bartter’s and Gitelman’s Syndromes Two human genetic tubulopathies, Bartter’s (BS) and Gitelman’s (GS) syndromes, have normo/hypotension and absent cardiac remodeling despite their apparent angiotensin system (RAS) activation. This seeming contradiction has led to an extensive investigation of BSGS patients, the result of which is that BSGS represents a mirror image of hypertension. BSGS’s unique set of properties has then permitted their use as a human model to probe and characterize RAS system pathways and oxidative stress in cardiovascular and renal remodeling and pathophysiology. This review details the results using GSBS patients that provide a deeper understanding of Ang II signaling and its associated oxidants/oxidative stress in humans. By providing a more complete and complex picture of cardiovascular and renal remodeling pathways and processes, studies of GSBS can inform the identification and selection of new targets and therapies to treat these and other oxidant-related disorders. Introduction Bartter's (BS) and Gitelman's (GS) syndromes are two rare human genetic tubulopathies. They are clinically characterized by hypokalemia, normo/hypotension, metabolic alkalosis, hypercalciuria or normocalciuria (in BS) or hypocalciuria (in GS), muscle weakness, polyuria, failure to thrive, polydipsia and other syndrome-specific side effects [1,2]. These syndromes are the result of impaired potassium reabsorption in the kidney, specifically in the thick ascending limb (TAL) and the distal convoluted tubule (DCT), due to mutations of genes encoding the ion channels and cotransporter involved in nephron's electrolyte trafficking [1,2]. This altered electrolyte trafficking gives rise to renin-angiotensin system (RAS) overactivation. However, strikingly, these patients have normo-or hypotension and do not exhibit the cardiovascular-renal remodeling classically associated with arterial hypertension [3,4]. Furthermore, in addition to the activation of multiple pathways in RAS, like those found in hypertension, they unexpectedly exhibit increased antioxidant defenses, as well. Given these seeming paradoxes, BSGS patients represent an extraordinary opportunity in a human model to uncover and assess potential interventions of RAS-and oxidative stress (OxSt)-related cardiovascular and renal remodeling pathways [3,4]. Nature has provided in BSGS a model showing a blunting and/or reversal of the deleterious effect of increased Ang II-based RAS and ROS signaling (hypertension, cardiac remodeling, etc.). The results from BSGS patients should be useful for identifying targets using BSGS results as a guide to explore potential treatments in patients with hypertension, cardiac remodeling, etc., that mimic the positive effects noted in BSGS. In other words, Antioxidants 2023, 12, 811 2 of 8 understanding why BSGS patients do not develop hypertension and its long-term complications and do not present OxSt despite high Ang II and the activation of RAS may give insights into the molecular bases of hypertension and its complications and might provide clues for targets of therapy in hypertension and its long-term complications such as cardiovascular and renal remodeling. Gitelman's and Bartter's Syndromes Gitelman's syndrome was first described in 1966 as a familial disorder characterized by electrolyte disturbances with concomitant hypomagnesemia and hypokalemia [5]. GS's clinical features are like those induced by the side effect of thiazide' diuretics therapy, with salt craving, hypotension and muscle cramps. The prevalence of the disease is 1/40,000, making it one of the most common tubulopathies [2]. GS is often quite difficult to diagnose as it requires a high level of clinical suspicion. In fact, a major portion of the population might be affected without knowing [6]. Several mutations have been found in the SCL12A3 gene which encodes for the sodium chloride cotransporter (NCC) in the DCT (Table 1) [7]. The prevalence of BS is 1/1,000,000 and the clinical features to a large extent overlap those of GS, e.g., muscle weakness, anorexia, polydipsia, polyuria, failure to thrive and growth retardation with salt-wasting [1]. A specific feature that distinguishes BS from GS is its very early presentation with growth retardation and failure to thrive [1]. Moreover, during pregnancy, it can produce polyhydramnios and induce premature delivery. Hypercalciuria is another consequence of the excessive sodium wasted, which increases the risk of nephrocalcinosis. There are multiple subtypes of Bartter's syndrome ( Table 1) that are related to the specific gene affected [1] (Table 1). ROS, Oxidative Stress and Renin Angiotensin System Activity Oxidative stress is the loss of intracellular redox homeostasis when the equilibrium between antioxidant and pro-oxidant factors is altered [8]. Reactive oxygen species (ROS) usually refers to free radicals and other non-radical intermediates which quickly react with all the surrounding molecules [9,10]. Superoxide (O 2 − •) is among the most prominent ROS species, as O 2 − •, given its relative stability, is able to cross membranes, where it not only acts as a signaling molecule but undergoes catalytic decomposition to •OH, an extremely reactive radical [11,12]. Other reactive radical compounds include radical nitric oxide (NO•) and peroxynitrites (ONOO−) [11]. ROS overproduction occurs mainly in mitochondria to produce ROS-driven damage to cellular lipid, proteins and DNA. In addition to direct damage, ROS alter cellular signaling to activate the inflammatory system as well as causing programmed cell death [13]. Another example is that ROS generated by vascular NOX modulates nitric oxide (NO) bioavailability, thereby regulating blood pressure [14]. Among ROS protective mechanisms, hypoxia inducing factors (HIF) are particularly important. HIF regulate the NOS system, which in turn stimulates HO-1 and Cyclooxygenase 2 (COX2) [20] and the nuclear factor E2 associated factor (Nrf2) system to deal with the ROS overproduction [21]. HO-1 has a very low basal expression but increases rapidly upon oxidative stress. In addition, HO-1 mediates the production of carbon monoxide CO, a known vasodilator, thereby contributing to the regulation of vascular tone, blood pressure and endothelial function [18]. Oxidative stress is a key factor in the pathological progress of atherosclerosis, which underlies cardiovascular diseases such as ischemic heart diseases, stroke and peripheral arterial diseases. Oxidative stress drives endothelial dysfunction by altering endothelial signal transduction and redox-regulated transcription factors. This then increases vascular endothelial permeability and favors leukocyte adhesion [22]. The increase in endothelial permeability allows LDL access to the vessel wall, where they are oxidized to oxLDL [23]; oxLDL are then engulfed by macrophages, becoming foam cells. The presence of foam cells alongside elevated levels of ROS and reduced NO levels promote muscle hyperplasia and dysregulated intima structures such as vascular calcifications. Critically, endothelial damage caused by free radicals results in a vicious circle as damage to the endothelium promotes the further production of free radicals [24,25]. RAS participates in the regulation of a plethora of systems, including blood pressure, fluid and electrolyte balance and systemic vascular resistance. Ang II is a major RAS signaling component whose Janus-faced activity can promote either vasoconstriction, inflammation, fibrosis and cellular growth, or vasodilation, insulin sensitivity, anti-remodeling, and anti-atherogenesis effects [15]. The RAS signaling cascade results from the binding of Ang II with its cell-surface Ang II subtype-1 or subtype-2 receptors (AT1R, AT2R). AT1R receptors are members of the seven transmembrane-domain G protein-coupled receptors superfamily, and their binding promotes the coupling of G proteins (G q /G 11 and/or Gi/G 0 ) to the C-terminal domain of the receptor, stimulating many intracellular signaling pathways for PLC, the activation of Ca 2+ channels, PLA 2 , adenylate cyclase, MAP kinases, the JAK-STAT pathway and NADPH oxidase (NOX), amongst others [15,17]. Given the critical role of Ang II signaling in the regulation of vascular tone, altered Ang II signaling plays a central role in cardiovascular-renal pathophysiology. Persistent activation of the RAS in healthy individuals leads to hypertension and target organ damage. This damage results from both short and long-term effects of Ang II, thus making the short and long-term signaling pathways of Ang II important subjects of investigation to understand, at a molecular level, the mechanisms that cause hypertension and cardiovascular-renal remodeling [15]. For example, Ang II signaling in chronic kidney disease (CKD) produces alterations in the vasculature as well as in the heart, such as cardiovascular remodeling and hypertension [18] (Figure 1A) [25]. Endothelial dysfunction and atherosclerotic changes are prominent features of CKD even at the very early stages [18]. Endothelial and vascular smooth muscle cells (VSMCs) crosstalk in the context of increased ROS, IL-1 and decreased levels of NO, which characterize CKD and promote VSMCs migration toward intima, causing hyperplasia vascular calcification and inducing arterial stiffness [26,27]. These mediators can also affect the function of epithelial cells of renal tubules and promote epithelial-to-mesenchymal cells transition (EMT), causing kidney fibrosis [14,27].Overall, Antioxidants 2023, 12, 811 4 of 8 the RAS, and Ang II as its major effector, are involved in cardiovascular-renal remodeling in CKD by driving oxidative stress and inflammation [17]. tension [18] (Figure 1A) [25]. Endothelial dysfunction and atherosclerotic changes are prominent features of CKD even at the very early stages [18]. Endothelial and vascular smooth muscle cells (VSMCs) crosstalk in the context of increased ROS, IL-1 and decreased levels of NO, which characterize CKD and promote VSMCs migration toward intima, causing hyperplasia vascular calcification and inducing arterial stiffness [26,27]. These mediators can also affect the function of epithelial cells of renal tubules and promote epithelial-to-mesenchymal cells transition (EMT), causing kidney fibrosis [14,27].Overall, the RAS, and Ang II as its major effector, are involved in cardiovascular-renal remodeling in CKD by driving oxidative stress and inflammation [17]. [3,4,15]. Panel (A). Angiotensin II signaling in chronic kidney disease and hypertension promotes through oxidative stress vasoconstriction, vascular remodeling and insulin resistance. Ang II, angiotensin II; AT1R, angiotensin II receptor 1; PLC β, phospholipase C β; PIP2, phosphatidylinositol diphosphate; DAG, diacylglycerol; PKC, protein kinase C; eNOS, endothelial nitric oxidase; NOX, NAD(P)H oxidase, ROCK, Rho Kinase; MYPT, myosin phosphatase protein target. Panel (B). Angiotensin II signaling in GSBS is blunted at post-receptor level, thereby promoting vasodilation and insulin sensitivity through RGS-2 activity. RGS, regulators of G-protein signaling; HO-1, heme-oxygenase 1; NO, nitric oxide. Figure 1. Angiotensin II signaling in chronic kidney disease and hypertension versus angiotensin II signaling in GSBS A frequent CKD comorbidity is atrial fibrillation (AF), a tachyarrhythmia with irregular and rapid heart rate, which occurs alongside heart failure with preserved ejection fraction and poorly controlled arterial hypertension [28]. There are multiple mechanisms through which ROS can potentially stimulate AF, by causing increased ion leak throughout mycardiocytes [26], as well as by stimulating collagen deposition and fibrosis that interferes with the propagation of the electric impulse [29]. Ang II increases Rho kinase (ROCK) activity, leading to elevated myosin phosphatase target protein subunit-1 (MYPT-1) phosphorylation, and to the increased expression of Connexin 40, an integral membrane protein of heart gap cells junction, which enhances atrial vulnerability to electrical disturbances [30,31]. A frequent CKD comorbidity is atrial fibrillation (AF), a tachyarrhythmia with irregular and rapid heart rate, which occurs alongside heart failure with preserved ejection fraction and poorly controlled arterial hypertension [28]. There are multiple mechanisms through which ROS can potentially stimulate AF, by causing increased ion leak throughout mycardiocytes [26], as well as by stimulating collagen deposition and fibrosis that interferes with the propagation of the electric impulse [29]. Ang II increases Rho kinase (ROCK) activity, leading to elevated myosin phosphatase target protein subunit-1 (MYPT-1) phosphorylation, and to the increased expression of Connexin 40, an integral membrane protein of heart gap cells junction, which enhances atrial vulnerability to electrical disturbances [30,31]. RAS and ROS in Gitelman's and Bartter's Syndrome Patients The overactivation of the RAS with concomitant reduced peripheral resistance, normal to low blood pressure and the absence of cardiovascular/renal remodeling, are the counterintuitive clinical features of GSBS patients ( Figure 1B) [3,4]. Despite having elevated levels of Ang II, similar to those found in hypertensive subjects, GSBS patients have decreased α subunit of Gq protein (Gαq) expression, which as a result promotes intracellular Ca 2+ release and PKC activation [3,32]. For example, in GSBS the decreased expression of Gαq influences the interplay between GDP dissociation and GTP binding; the subsequent PLCβ activation, which in turn elevates diacylglycerol (DAG) and IP3 production, leads to the increased release of intracellular Ca 2+ [3,33]. The altered AT1R downstream signaling in GSBS is also modulated by the regulators of G-protein signaling (RGS) [34]. These proteins are critical for the G protein-coupled receptor (GPCR) activity because they control the signaling via GPCR by regulating a variety of effector proteins [3,35]. RGS can inhibit GPCR signaling in several ways: by accelerating GTP hydrolysis (GTPase activating protein-GAP) that turns off the GPCR signal; by decreasing GPCR sensitivity to its agonists; and by boosting signal decay after the GPCR agonist removal. Interestingly, and again in contrast to hypertensives who display decreased RGS-2 expression [36], GSBS patients have increased gene and protein expressions of RGS-2 [34]. GSBS patients' increased expression of RGS-2 reduces the expression of the Gα subunit, resulting in reduced PLCβ activity, IP3 and DAG [3,34,37]. Therefore, Ca 2+ release is reduced and less Ca 2+ -calmodulin (Ca 2+ /CaM) complex is formed. In addition, DAG reduces PKC activity, with consequent increased eNOS expression and NO production, resulting in reduced vascular tone [3] (Figure 1B). These intracellular pathways alter muscle contraction and peripheral resistance by reducing the phosphorylation of the regulatory chain of myosin II [3]. The activation of the G-protein RhoA and its effector Rho kinase (ROCK) regulates the phosphorylation state of the MYPT-1, which is the regulatory subunit of the myosin light chain phosphatase (MLCP) [35,38,39]. The RhoA/ROCK pathway in GSBS is downregulated, which reduces phosphorylated MLC and produces vasodilation [3,40,41]. In contrast, the RhoA/ROCK pathway is upregulated in hypertension, where it induces vasoconstriction [3,41]. In addition, in GSBS the expression of RhoA regulators is altered upstream as they have reduced expression of Rho guanine nucleotide exchanger factors (RhoGEFs) in association with increased guanine nucleotide dissociation inhibitor (RhoGDI) that inhibits the dissociation of GDP from GDI [3,40]. This is in contrast to the increased expression of p63RhoGEF and p115RhoGEF found in hypertensive patients [40,41]. In this way, when RhoA is in its inactive form, the upstream signaling for ROCK results is blunted. A further feature of GSBS is the increased eNOS generated vasodilator NO, despite the increased activation of RAS, but consistent with reduced oxidative stress [3]. eNOS is known to be negatively regulated by PKC, a kinase activated via G protein signaling. However, as noted, GSBS patients' G protein signaling is altered, which reduces activated PKC, thereby increasing eNOS activation with ensuing increased NO production [3] ( Figure 1B). Moreover, the response of NOXs to Ang II is reduced in GSBS patients. The stimulation of monocytes isolated from GSBS patients with Ang II in vitro shows reduced p22 phox gene expression and TGFβ activity, which demonstrates reduced oxidative pathway activity in these patients [3,42]. Altogether, the repeated findings in GSBS patients that are the inverse of the pathway responses in hypertensives show that GSBS patients should be considered the mirror image of hypertension. Therefore, these patients provide a very useful human model to explore, identify and interrogate the roles played by RAS and oxidative stress in cardiovascular pathology and renal remodeling ( Table 2). Conclusions The study of GSBS patients has identified multiple oxidant-related systems which suggest several potential therapeutic targets to reduce and/or block Ang II-related as well as other RAS-and ROS-related processes. As detailed above, the complex array of altered pathways and ROS-related effects found in GSBS clearly must combine to block the pathological effects typically associated with an elevated Ang II. The myriad of oxidants, as well as changes in their levels, alongside the multiple pathways, organelles and organ systems involved found in GSBS patients, provides a rationale for why treatments aimed solely at "oxidants" have proven mostly unsuccessful on clinical grounds [43,44]. However, blocking multiple Ang II dependent signals, such as via blocking AT1R and ROCK activity, for example, does lead to the improvement of cardiovascular-renal remodeling, in part via a reduction of oxidative stress [18,45,46]. In summary, GSBS as a human model obverse that of hypertension has provided and will continue to provide important clues as to new targets, as well as improved treatment regimens in hypertension and its long term complications, aimed at Ang II and oxidative stress mediated cardiovascular renal remodeling.	2023-03-29T15:12:54.729Z	2023-03-26T00:00:00.000	{ "year": 2023, "sha1": "e3abfd80d144cab7853b29bd14b10008c4b11d61", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2076-3921/12/4/811/pdf?version=1679826052", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "e9dfd7c40805ba332e7c08f8270b95c51ac03b0d", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
252600737	pes2o/s2orc	v3-fos-license	Climate change may outpace current wheat breeding yield improvements in North America Variety adaptation to future climate for wheat is important but lacks comprehensive understanding. Here, we evaluate genetic advancement under current and future climate using a dataset of wheat breeding nurseries in North America during 1960-2018. Results show that yields declined by 3.6% per 1 °C warming for advanced winter wheat breeding lines, compared with −5.5% for the check variety, indicating a superior climate-resilience. However, advanced spring wheat breeding lines showed a 7.5% yield reduction per 1 °C warming, which is more sensitive than a 7.1% reduction for the check variety, indicating climate resilience is not improved and may even decline for spring wheat. Under future climate of SSP scenarios, yields of winter and spring wheat exhibit declining trends even with advanced breeding lines, suggesting future climate warming could outpace the yield gains from current breeding progress. Our study highlights that the adaptation progress following the current wheat breeding strategies is challenging. Specific comments Supplementary Table S4: What is the physiological basis for using different Tbase temperatures for different growth stages? It was not clear whether the same check variety was used for winter wheat as well as spring wheat. L30: by 1.9% less yield losses --> with 1.9% less yield loss L32: However, additional 0.5% yields were lost per 1oC warming for elite spring wheat --> You mean on top of the 1.9% yield loss? L82 Figure 1a: (000 ha) --> (1000 ha) L82 Figure 1b and 1c: ton/ha--> you mean Metric ton /ha)? L125: are slightly difference --> are slightly different L146-147: Finally, we apply the temperature thresholds of spring wheat in Supplementary Table S4 to winter wheat, and generate a very similar results --> It was not clear how this was done (matching by stages?) L164: for Kharkof --> Not mentioned anywhere else in the paper L165: Marquis --> Not mentioned anywhere else in the paper L225-226: The sentence is not clear L229-231: Yield was estimated to decline by around 4.0-6.7% for winter wheat and 15.6-17.4% for spring wheat per each 100oCd increase in EDD --> Yield is log-transformed before regression. Are you sure the regression coefficient represents proportional change? L293: It is very difficult to mimic those rare new normal weather events in the traditional breeding approach --> what you mean by 'rare new normal weather events' ? L296: and more variety testing in multiple environments to prevent new varieties from leaving behind the future changes in climate --> How multiple environments can address future changes in climate? L372: as percentage changes in yield --> Is this true with your log-transformed yield data in regression? L374: resampling (with replacement) --> what is the resampling size L377: multicollinearity among predictors--> How about time (i.e. year) (Eq.5)? Reviewer #2: Remarks to the Author: There is a lot of value to this study in terms of the impact of climate on wheat. I think there are some opportunities to explain the results better and provide a better "blueprint" for wheat breeders. Some of the approaches mentioned for breeders in the discussion section are outdated. My comments are below: Line 64: Overall the authors do a poor job of providing a blueprint to "guide future breeding programs". Many of their suggestion in the discussion are outdated and oversimplify a complex breeding target. Line 65: Use of 'severely' is an overstatement and not really true -plenty of studies in this area. Overall there needs to be more intro on what impacts wheat yield, such as vernalization requirement in winter wheat vs spring wheat. Currently little to no information is provided to the readers. Picking one and showing it can be very misleading and uninformative. How has the yield of the CK cultivar changed overtime? Previous studies have showed it increasing in yield. Showing % yield change instead of just yield would also be good so that it can be comparable to figure 3. Currently it is very hard to compare the two. Line 118-120: So how much of the "less yield loss" can be attributed to each of these two components? Line 124-126: Often times in these nurseries, locally adapted lines will be higher yielding. For example, Colorado breeding lines will perform better in Colorado compared to other breeding lines. I would imagine that this would be even more likely in winter wheat compared to spring wheat due to specific vernalization requirements for different sites. I think this idea of local adaptation could be influencing the results you are seeing here for winter wheat and the better performance to climate and compared to spring wheat. Also, higher yielding lines will often be more negatively impacted by heat stress compared to average or low yielding lines -even though they will still be higher yielding. Greater reduction vs. final yield. Do the authors have any thoughts or discussion on local adaptation? Line 125: Grammar Line 133-134: Can the authors speculate as to why this is happening? Is it changes in phenology such as heading date or maturity date? Figure 3: I think the authors need to do a better job of projecting the next 80 years in the context of what has happened the last 60 years in their dataset. Though not shown (other than the Kansas example in figure 1) there has been a continuous increase in wheat yield from 1960 to 2020 and then based on this figure, we will see a rapid decline in this progress starting immediately. It's a little confusing and the lack of synergy between figure 1 and 3 does not help. 190-194: Why do the authors focus on MYG instead of HYG? Breeders do not release MYG. Also, there is a misuse of the word varieties. These are mostly breeding lines -some were released as varieties, but not the majority. Varieties are only those that are commercially available to a farmer. 225-228: Grammar issue here Line 235-237: Breeding is not resulting in yield loss, climate is. This is misleading. 246-248: This has also been observed in CIMMYT Spring Wheat 286-287: With new technologies such as speed breeding, doubled-haploids, and genomic selection, variety development is done much more rapidly. These should be discusses if authors are making this claim. Line 311-313: This is an oversimplification of the complexity of breeding -most methods now focus on a whole genome/genomic selection approaches, as marker assisted selection for a few genes will never be effective for a quantitative trait like heat tolerance. 341-344: What is the precedent for using these percentiles? 357-358: A lot of first person writing -i.e. We 358-360: Planting and harvest dates are likely available for most site years for this dataset and they can vary widely -why did the authors choose to use a previously published average? Seems like an oversimplification Reviewer #3: Remarks to the Author: Thank you for preparing this interesting manuscript. I enjoyed reading it a lot. I agree that climate change impact modelling studies often point to the possibility of adapting the cultivars without discussing how effective such adaptation might be or whether the cultivars for that even exists yet. So I think your manuscript is an important contribution to that discussion. I have mostly minor comments and feedback on specific sentences or how to present the results and some requests for more information, for instance about the yield data used. Please find them in the attached annotated manuscript. I think that the discussion section could be shortened and improved. There are a lot of descriptions of results which should be removed or moved to only focus on the discussion of the results. I would avoid referring to specific figures and tables in this section. I appreciate that you want to keep the results section concise but it should not come at the cost of a longer discussion section. Perhaps work with sub-headings. I am a bit concerned about the decision to use a published dataset for planting and harvest dates when you seem to have reported planting and harvest dates from the reports. There might be quite a deviation and you could not consider changes to these dates over time which could influence your results. Please explain your decision on this and potential bias introduced. I am happy to review a revised version of the manuscript should it be required. Response letter to reviewers We greatly appreciate these insightful comments and suggestions to our manuscript. We revised as you request and provided more analyses in the version, especially for CO2 fertilization effect, potential uncertainties due to the lack of crop calendar data, and more discussions on new breeding technologies. Please check our point-by-point relies and associated revision in the new version. All replies have been marked in blue, and our revision has be highlighted by yellow in the main text. General Comments This paper provides a comprehensive analysis on the impact of climate change on winter and spring wheat yield using a large historic dataset and fixed-effect panel data regression models. Analysis using growing degreed days (GDD), extreme degree days (EDD) and freezing degree days (FDD) are appropriate, and results provide valuable insights on the impact of these variables on wheat grain yield in current and future climates. Reply: We greatly appreciate your suggestions. We have revised based on your comments, and provided an analysis for CO2 fertilization effect. Please check our point-by-point rely: Weaknesses The fixed-effect panel data regression model (Eq. 5) has a time (i.e. year) component, but the paper did not report any results on the site-specific time trend (linear and quadratic coefficients). Reply: In panel data regression model, the values of site, site-year and site-year 2 are "dummy variables to remove non-climate factors" (lines 360). We attached a XLSX file to show reviewers the full regression coefficients. Please check the values in Full_RegCoef.xlsx. In our manuscript, we focused on how changes in climate affect yields. Therefore, we only reported the regression coefficients of climate variables in the main text. The paper grouped the new experiment lines into three categories of high-yielding (HYG), median-yielding (MYG), and low-yielding (LYG) genotypes. But it was not clear which varieties were used in each year during 1960-2018 and it was not clear how yield gain through time (i.e year) was determined. Reply: Sorry about the confusion. The advanced breeding lines entered into the nursery programs studied each year are different than the previous year. Therefore, we did not fix a certain breeding line. Instead, we used a relative term to quantify various yield levels of HYG, MYG and LYG. More specifically, HYG is the 97.5% percentile grain yield for advanced breeding lines each year, MYG is 50% percentile, and LYG is 2.5% percentile. To address this concern, we added the following sentence: "On the other hand, the advanced breeding lines entered each year were different from the previous year, which reflects the ongoing nature of wheat breeding itself over the study period." in lines 82-84. If the temperature stress can be completely canceled out by variety replacement, yield sensitivity of advanced breeding lines to climate should be lower than results based on CK yield data series, and vice versa. We added the following sentence: " Therefore, the difference in yield sensitivity (advanced breeding lines vs. check variety) can be defined as effectiveness of the current variety development under climate change, providing a footprint of the actual wheat breeding effort." (lines 85-87) CO2 concentration was not included in the regression models and CO2 effect on yield was not included in the analysis and discussion. Reply: We appreciate the reviewer bringing up this point and agree that it is important to address. Earlier study (Lobell and Field, 2007) has found that CO2 effect is difficult to quantify based on a statistical approach using historical crop statistics because the relatively small importance of year-to-year CO2 changes for yield variability results in the high uncertain estimates of CO2 responses from such approach. To demonstrate this, we followed the approach of Lobell and Field (2007) and included CO2 item in a panel data regression model (the Table R1 in next page). Similar to Lobell and Field (2007), for the US we find that: (1) The regression coefficient of CO2 is negative and insignificant in many cases, which is physiologically irrational with the CO2 fertilization effects. Our estimate is the same with Fig. 6 in Lobell and Field (2007) paper, in which they also found negative regression in US wheat. We believe this is because the year-to-year CO2 variability is much smaller than climate impact, and very difficult to have clear effects detected based on statistical model. (2) The regression coefficients of other climate variables are not affected by inclusion of CO2, which reflects robustness of our estimation for other climate variables. The above points indicate that estimating a CO2 fertilization effect directly in a regression framework is unlikely to be accurate. As an alternative approach, we instead take advantage of the study design of comparing HYG, MYG, and LYG yields to a check variety. We have rewritten our results to present our findings as the difference between pairs of data accordingly. Presented in this way, the results on spring and wheat yield gains of variety adaptation quantified is robust regardless of CO2 levels because CO2 levels within a year are constant across varieties within each growing season. We added the sentences: " Finally, we note that studies based on panel data model do not often consider CO2 fertilizations 44,45 . This is because the relatively small importance of year-to-year CO2 changes for yield variability results in the highly uncertain estimates of CO2 responses from such approach 47 . Additionally, results on spring and wheat yield gains of variety adaptation quantified is robust regardless of CO2 levels because our analysis provides pairs of data between advanced breeding lines and constant check varieties in each growing season. We encourage future field investigations and variety testing on different CO2 levels." in lines 373-380. Lobell, D., Field, C., Estimation of the carbon dioxide (CO2) fertilization effect using growth rate anomalies of CO2 and crop yields since 1961. Glob. Change Biol. 14, 39-45 (2008). To show how much the yield sensitivity of our breeding nurseries can represent the actual sensitivity for farmer's yields (that is, the real wheat releases), we compare the yield sensitivity to 1 o C warming based on our data with the results based on county-level yield statistics. We rewrote the sentence: "when the above statistical analysis was repeated but used the county-level yield statistics, the yield responses of MYG to 1 o C warming derived here were close to the estimation from county-level yield statistics ( Supplementary Fig. S2). This reflected that the response of MYG can approximate and represent the yield sensitivity in farmer's fields at regional scale. " in lines 151-155. Reply: Earlier studies showed that the critical temperature thresholds vary with growth phases of wheat. Supplementary Please find these references in the caption of supplementary L32: However, additional 0.5% yields were lost per 1oC warming for elite spring wheat --> You mean on top of the 1.9% yield loss? Reply: The "1.9% less yield losses" is for winter wheat and "0.5% yield" is for spring wheat. However, as request by reviewer#3, the sentence has been removed. We now show yield decline for advanced breeding lines and check variety directly: "Results shows that yields were declined by 3.6%/ o C for elite winter wheat breeding lines, compared with -5.4%/ o C when variety is held constant, reflecting a superior climate-resilience. However, advanced spring wheat breeding lines demonstrate 7.5% yield reduction per 1 o C warming. That is more sensitive than the value of -7.1%/ o C with constant variety planted, indicating an undermined climate-resilience for spring wheat.". Please check lines 32-37. L146-147: Finally, we apply the temperature thresholds of spring wheat in Supplementary Table S4 to winter wheat, and generate a very similar results --> It was not clear how this was done (matching by stages?) Reply: Sorry about the confusion. As you said, we match growth phases for heading to maturity. Before that stage, we believe some genetic difference still exists as their genetic basis is quite different between winter and spring wheat (i.e. winter wheat needs to experience winter but spring wheat does not). To address the confusion, we rewrote the sentence: "Finally, we apply the temperature thresholds of spring wheat in Supplementary Table S4 to winter wheat from heading to maturity, and generate a very similar results for winter wheat yield response to 1 o C warming" (lines 161-163). L164: for Kharkof --> Not mentioned anywhere else in the paper Reply: Sorry about the confusion. To address this, we rewrote the sentence: " The variety Kharkof was used as the long-term winter wheat check, and the variety Marquis as the long-term spring wheat check. The two check varieties were planted in each site throughout the study period, which could be viewed as yield variability without variety replacement. " in lines 79-82 and lines 324-325. L165: Marquis --> Not mentioned anywhere else in the paper Reply: Sorry about the confusion. To address this, we rewrote the sentence: " The variety Kharkof was used as the long-term winter wheat check, and the variety Marquis as the long-term spring wheat check. The two check varieties were planted in each site throughout the study period, which could be viewed as yield variability without variety replacement. " in lines 79-82 and lines 324-325. L225-226: The sentence is not clear Reply: Sorry about the confusion. We rewrote the sentence: " Warming poses a more harmful effect on yield of spring wheat breeding, resulting primarily from greater sensitivity to EDD of the advanced spring wheat breeding lines than CK." in lines 251-253. L229-231: Yield was estimated to decline by around 4.0-6.7% for winter wheat and 15.6-17.4% for spring wheat per each 100oCd increase in EDD --> Yield is log-transformed before regression. Are you sure the regression coefficient represents proportional change? Reply: Yes. The regression coefficients represent the percentage change in dependent variables for every one-unit increase in the independent variable. Please check the trivial example below: Suppose the fitted model is log(weight) = 2.14 + 0.00055 height The estimated coefficient for height is 0.00055, so we would say that an increase of one unit in height is associated with a 100 × ( 0.00055 − 1) ≈ 0.055 percent change in weight. To address this concern, we added the following sentence: "The regression coefficients determine the wheat yield response as percentage changes in yields for each one-unit increase in climate variables." in lines 364-365. L293: It is very difficult to mimic those rare new normal weather events in the traditional breeding approach --> what you mean by 'rare new normal weather events'? Reply: In breeding programs, breeders still rely on phenotypic selection approach. In the actual field conditions, it is very difficult to create an environment with very high temperature (i.e. the "rare new normal weather events"). Therefore, it will be very difficult for breeders to select the breeding lines with potential heat tolerance using such a traditional breeding approach. As request by reviewer#2, we have rewritten the part and discuss more state-of-the-art breeding technologies, which reads" we note that most breeding programs relied heavily on phenotypic selection approach during the past decades represented by the nursery programs used for this analysis. This suggests that new breeding technologies are required to be implement now, such as speed breeding 35 , doubled-haploids 36 , and genomic selection 37 may have great value for accelerating the development of new varieties, particularly if major genetic determinants for the underlying process of heat stress adaptation are identified 38 . The exact value of these new breeding innovations as they pertain to heat tolerance should be clarified within the next decade. Continued assessment of new advanced lines should be evaluated in multiple environments to be exposed to local weather extremes which will opportunistically identify improved tolerance to temperature extremes." in lines 289-299. L296: and more variety testing in multiple environments to prevent new varieties from leaving behind the future changes in climate --> How multiple environments can address future changes in climate? Reply: To us, we need to test the advanced breeding lines performance under various heat stress environments vs. normal temperature regime (i.e. the "multiple environments"), and identify the major genetic determinants for the underlying process of heat stress. As request by reviewer#2, we have rewritten the part and discuss more state-of-the-art breeding technologies, which reads" we note that most breeding programs relied heavily on phenotypic selection approach during the past decades represented by the nursery programs used for this analysis. This suggests that new breeding technologies are required to be implement now, such as speed breeding 35 , doubled-haploids 36 , and genomic selection 37 may have great value for accelerating the development of new varieties, particularly if major genetic determinants for the underlying process of heat stress adaptation are identified 38 . The exact value of these new breeding innovations as they pertain to heat tolerance should be clarified within the next decade. Continued assessment of new advanced lines should be evaluated in multiple environments to be exposed to local weather extremes which will opportunistically identify improved tolerance to temperature extremes." in lines 289-299. L372: as percentage changes in yield --> Is this true with your log-transformed yield data in regression? Reply: Yes. The regression coefficients represent the percentage change in dependent variables for every one-unit increase in the independent variable. Please check the trivial example below: Suppose the fitted model is log(weight) = 2.14 + 0.00055 height The estimated coefficient for height is 0.00055, so we would say that an increase of one unit in height is associated with a 100 × ( 0.00055 − 1) ≈ 0.055 percent change in weight. To address this concern, we added the following sentence: "The regression coefficients determine the wheat yield response as percentage changes in yields for each one-unit increase in climate variables." in lines 364-365. L374: resampling (with replacement) --> what is the resampling size Reply: This is the bootstrapping method that often used to determine the uncertainty of regression model in climate impact studies (Lobell et al., 2012). Bootstrapping is a method of resample. The idea is to use the observed sample to estimate the population distribution. Then samples can be drawn from the estimated population and the sampling distribution of any type of estimator can itself be estimated. For example, the observed sample is 1 1.1 2 2.2 3 2.9 4 3.5 . The first column is the independent data series and the second column is the dependent data series. We have a regression equation, y=0.79x+0.45. Then we draw randomly by resampling with replacement. The first resample might be 3 2.9 2 2.2 3 2.9 1 1.1 , and the second might be There is a lot of value to this study in terms of the impact of climate on wheat. I think there are some opportunities to explain the results better and provide a better "blueprint" for wheat breeders. Some of the approaches mentioned for breeders in the discussion section are outdated. My comments are below: Reply: We greatly appreciate your positive comments. We have revised, and provides more discussions on the new breeding technologies and how to develop heat tolerance properties in an effective manner. Please check our point-by-point rely. Line 64: Overall the authors do a poor job of providing a blueprint to "guide future breeding programs". Many of their suggestion in the discussion are outdated and oversimplify a complex breeding target. Reply: Based on this suggestion, we have rewritten the part and discuss more state-of-the-art breeding technologies, which reads" we note that most breeding programs relied heavily on phenotypic selection approach during the past decades represented by the nursery programs used for this analysis. This suggests that new breeding technologies are required to be implement now, such as speed breeding, doubled-haploids, and genomic selection may have great value for accelerating the development of new varieties, particularly if major genetic determinants for the underlying process of heat stress adaptation are identified. The exact value of these new breeding innovations as they pertain to heat tolerance should be clarified within the next decade. Continued assessment of new advanced lines should be evaluated in multiple environments to be exposed to local weather extremes which will opportunistically identify improved tolerance to temperature extremes." in lines 289-299. Line 65: Use of 'severely' is an overstatement and not really true -plenty of studies in this area. Reply: Earlier studies focus on the comparison between yield itself of new and check varieties. Here, we focus on the yield sensitivities to climate with and without genetic advancement, which has not been studied much based on our knowledge. We revised as request: "To date, the relevant understanding remains lacking because of the scarcity of long-term field observations enabling comparison between yield sensitivities to climate with and without genetic advancement." in lines 68-71. Overall there needs to be more intro on what impacts wheat yield, such as vernalization requirement in winter wheat vs spring wheat. Currently little to no information is provided to the readers. Reply: Thanks for this suggestion. We introduced several important references to processes of warming that reduces wheat yields: "Many plant species are negatively affected by high temperature extremes, especially during floral stage 11 , and often have a shortened grain filling period 12 , less biomass accumulation 10 and consequently lower grain yields 3,4,5,6 " in lines of 55-58. Besides, we also stated that, to understand the different response of winter and spring wheat, "Such differences in responses show the need to understand further the underlying genetic process involved, such as specific vernalization requirements and variety responses to heat/cold stress between winter wheat versus spring wheat. More control field experiments are needed to examine the genetic basis across winter and spring wheat genotypes in future investigations." in lines of 262-266. Reply: We agree with the concern. Indeed, many earlier studies have found increase in CK yield. From our data, the yield trend of CK variety is also increasing for winter and spring wheat over time ( Figure 1R below). However, we should note that the site number is not constant in each year. For example, for SRPN, there were 15 sites in 1960 but the number of sites is 24 in 2018. And some sites were not reported in some years and some new site was included in the breeding program. Therefore, the time trend of the average yield across all the sites not only include climate signal but also includes some information of changing site. So we believe this time trend for average yield cannot reflect yield response to climate. In our Fig. 2, the regression coefficients are estimated based on all data, so this is the demonstration of climate impact on yields. To address this concern, we picked two sites just for demonstrating how we determine CK, LYG, MYG and HYG yield series to readers. Besides, we have moved them to Supplementary Fig. S1 to avoid such misleading. Line 118-120: So how much of the "less yield loss" can be attributed to each of these two components? Reply: Sorry about the confusion. We rewrote the sentence " For example, with 1 o C warming, yield response is lowered by 1.6% to the associated increase in EDD for advanced breeding lines relative to check variety (MYG: -4.5% vs. CK: -6.1%). One-degree warming brings 0.71% greater yield benefits due to the associated decrease in FDD for MYG relative to CK (MYG: +1.58% vs. CK: +0.87%)." in lines 122-126. Line 124-126: Often times in these nurseries, locally adapted lines will be higher yielding. For example, Colorado breeding lines will perform better in Colorado compared to other breeding lines. I would imagine that this would be even more likely in winter wheat compared to spring wheat due to specific vernalization requirements for different sites. I think this idea of local adaptation could be influencing the results you are seeing here for winter wheat and the better performance to climate and compared to spring wheat. Also, higher yielding lines will often be more negatively impacted by heat stress compared to average or low yielding lines -even though they will still be higher yielding. Greater reduction vs. final yield. Do the authors have any thoughts or discussion on local adaptation? Reply: We agree that advanced breeding lines for a particular location will be better adapted there, i.e. the local-adaptation, but they represent only a fraction of the entries in any given year. The entries in the nurseries come from many geographically diverse breeding programs. Therefore, the advanced breeding lines tested in experiments has avoid local-adaptation issue. We have added the sentence " Each year the entries in the nurseries come from many geographically diverse breeding programs (The source of the breeding lines was listed as entries in each report of the NRPN, SRPN and HRSWURN)". in lines 321-322. In our results, it is not always the case for higher yielding genotypes to suffer more from heat injury than lower yielding genotypes. For example, the yield decline of MYG is less than CK for winter wheat, and the MYG is the higher yielding genotypes relative to the CK variety (Fig. 2). Reply: Sorry for the confusion. The yield change we project in Fig. 3 is not actual yield. It is the yield changes due to climate change under four SSPs. Actual yield is not predicable because the non-climate factors (changes in agronomy technologies, like fertilization and irrigation etc.) is not possible to predict. To address the confusion, we rewrote the figure captions " Figure 3. Model projection of wheat yield changes due to changes in climate of four categories of genotype under four SSPs, relative to the yield of CK at the baseline period." in lines 197-199. 190-194: Why do the authors focus on MYG instead of HYG? Breeders do not release MYG. Also, there is a misuse of the word varieties. These are mostly breeding lines -some were released as varieties, but not the majority. Varieties are only those that are commercially available to a farmer. Reply: Based on your suggestion, we revised Fig. 4 and put LYG, MYG and HYG together. We also compared the yield sensitivity of CK, LYG, MYG and HYG with county-level yield statistics. We found the "yield responses of MYG to 1 o C warming derived here were close to the estimation from county-level yield statistics (Supplemental Fig. S2)" (lines 151-155). We checked the whole manuscript to change varieties to breeding lines, as request. 225-228: Grammar issue here Reply: Revise as request: " Warming poses a more harmful effect on yield of spring wheat breeding, resulting primarily from greater sensitivity to EDD of the advanced spring wheat breeding lines than CK." in lines of 251-253. Line 235-237: Breeding is not resulting in yield loss, climate is. This is misleading. Reply: Thanks for pointing out this. We rewrote: "Spring wheat is projected to suffer more yield losses compared to winter wheat with future temperature increases" in lines 258-260. Reply: Appreciate your suggestion. Based on this suggestion, we have rewritten the part and discuss more state-of-the-art breeding technologies, which reads" we note that most breeding programs relied heavily on phenotypic selection approach during the past decades represented by the nursery programs used for this analysis. This suggests that new breeding technologies are required to be implement now, such as speed breeding 35 , doubled-haploids 36 , and genomic selection 37 may have great value for accelerating the development of new varieties, particularly if major genetic determinants for the underlying process of heat stress adaptation are identified 38 . The exact value of these new breeding innovations as they pertain to heat tolerance should be clarified within the next decade. Continued assessment of new advanced lines should be evaluated in multiple environments to be exposed to local weather extremes which will opportunistically identify improved tolerance to temperature extremes." in lines 289-299. Line 311-313: This is an oversimplification of the complexity of breeding -most methods now focus on a whole genome/genomic selection approaches, as marker assisted selection for a few genes will never be effective for a quantitative trait like heat tolerance. Reply: That is a very interesting point. To our knowledge, the reason that traditional genome selection is ineffective for heat tolerance breeding might reflect these traditional genome selection algorithms do not consider the underlying processes that environmental factors affect plant growth and development, and as such, have limited ability to capture Genotype × Environment interactions. Furthermore, robust phenotypic data for heat tolerance is required to validate a training set which continues to be difficult to generate. Recently, Dr. Tao Li, one of our co-authors, is working on integration of genomics with process-based crop modeling to overcome the disadvantages. The new technology is still under testing, for example, to predict phenology traits of rice varieties under different temperature regimes (Yang et al., 2022). However, this breeding technology is too young to implement right now. The exact value of these new breeding innovations requires to be evaluating and could be clarified within the next decade. We have rewritten the part and discuss more state-of-the-art breeding technologies, which reads" we note that most breeding programs relied heavily on phenotypic selection approach during the past decades represented by the nursery programs used for this analysis. This suggests that new breeding technologies are required to be implement now, such as speed breeding 35 , doubled-haploids 36 , and genomic selection 37 may have great value for accelerating the development of new varieties, particularly if major genetic determinants for the underlying process of heat stress adaptation are identified 38 . The exact value of these new breeding innovations as they pertain to heat tolerance should be clarified within the next decade. Continued assessment of new advanced lines should be evaluated in multiple environments to be exposed to local weather extremes which will opportunistically identify improved tolerance to temperature extremes." in lines 289-299. 341-344: What is the precedent for using these percentiles? Reply: Earlier studies often use the yield of five most productive breeding lines. But, here, we prefer a relative term to define yield levels. This is because the number of breeding lines to test is very different in each year. For example, there are only 7 entries in Colby in 1962 but the number is 50 in 2018. Therefore, we set the HYG as the 97.5% percentile of yields for advanced breeding lines and LYG as the 2.5% percentile. So, 95% entries could be included. 357-358: A lot of first person writing -i.e. We Reply: We have revised: " Growing season accumulation of GDD, EDD and FDD were calculated in each site-season pair." in lines 343-344. 358-360: Planting and harvest dates are likely available for most site years for this dataset and they can vary widely -why did the authors choose to use a previously published average? Seems like an oversimplification Reply: We agree with the concern. However, based on the data availability of USDA, only a small subset of sites recorded sowing and harvest dates. . We have revised: "Unfortunately, only a small subset of sites recorded phenology data, and therefore trial-specific growing season lengths could not be used without omitting a large fraction of the data. The average plant and harvest dates in each site therefore were used based on Sacks et al. for both winter and spring wheat. A fixed time window was often set for some temperature accumulation indices in climate impact studies, rather than the growing season of each individual year. This is because the latter would result in endogeneity in an analysis of FDD, GDD and EDD on yields (e.g. warmer season may not have higher value of GDD as shorter growing season). " in lines 345-352. Despite the endogeneity problem, we can still test whether this will influence our results. We randomly selected the planting and harvest dates in different years for a certain site based on the calendar range reported in Sacks et al. And we repeat the process for 1000 times, and did the regression analysis (Fig. R2). We found that: (1) The yield impact by 1 o C warming is slightly smaller than the results in Fig. 2. This reflects the endogeneity problem, and the degree-day cannot well reflect the actual temperature difference between years. (2) The major conclusion in the main text still holds: the yield decline of MYG is smaller than CK for winter wheat, the results is contrary for spring wheat. Therefore, we believe crop calendar does not influence our major conclusion. Thank you for preparing this interesting manuscript. I enjoyed reading it a lot. I agree that climate change impact modelling studies often point to the possibility of adapting the cultivars without discussing how effective such adaptation might be or whether the cultivars for that even exists yet. So I think your manuscript is an important contribution to that discussion. Reply: Great appreciation from us. We have moved some part of paragraphs to result part and make discussion shorter. Please check our point-by-point rely. I have mostly minor comments and feedback on specific sentences or how to present the results and some requests for more information, for instance about the yield data used. Please find them in the attached annotated manuscript. I think that the discussion section could be shortened and improved. There are a lot of descriptions of results which should be removed or moved to only focus on the discussion of the results. I would avoid referring to specific figures and tables in this section. I appreciate that you want to keep the results section concise but it should not come at the cost of a longer discussion section. Perhaps work with sub-headings. I am a bit concerned about the decision to use a published dataset for planting and harvest dates when you seem to have reported planting and harvest dates from the reports. There might be quite a deviation and you could not consider changes to these dates over time which could influence your results. Please explain your decision on this and potential bias introduced. Reply: We agree with the concern. However, based on the data availability of USDA, only a small subset of sites recorded sowing and harvest dates. . We have revised: "Unfortunately, only a small subset of sites recorded phenology data, and therefore trial-specific growing season lengths could not be used without omitting a large fraction of the data. The average plant and harvest dates in each site therefore were used based on Sacks et al. for both winter and spring wheat. A fixed time window was often set for some temperature accumulation indices in climate impact studies, rather than the growing season of each individual year. This is because the latter would result in endogeneity in an analysis of FDD, GDD and EDD on yields (e.g. warmer season may not have higher value of GDD as shorter growing season). " in lines 345-352. Despite the endogeneity problem, we can still test whether this will influence our results. We randomly selected the planting and harvest dates in different years for a certain site based on the calendar range reported in Sacks et al. And we repeat the process for 1000 times, and did the regression analysis (Fig. R2). We found that: (1) The yield impact by 1 o C warming is slightly smaller than the results in Fig. 2. This reflects the endogeneity problem, and the degree-day cannot well reflect the actual temperature difference between years. (2) The major conclusion in the main text still holds: the yield decline of MYG is smaller than CK for winter wheat, the results is contrary for spring wheat. Therefore, we believe crop calendar does not influence our major conclusion. Reply：This is because the reference is different between Fig. 3 and Fig. 4. In Fig. 3, the yield changes is relative to the CK varieties at the baseline climate. This is to show the potential yield trend induced by climate compared to the historical yield level of CK . Line 225: Unclear to me. Do you mean: "Comparison between winter and spring wheat shows that..." Reply: Sorry about the confusion. We have removed the sentence and now it is: " Warming poses a more harmful effect on yield of spring wheat breeding, resulting primarily from greater sensitivity to EDD of the advanced spring wheat breeding lines than CK. Spring wheat typically is at floral stage during a higher temperature cycle than winter wheat during its floral stage. " in lines 251-254. Line 225: add "the" Reply: We have removed the sentence and now it is: " Warming poses a more harmful effect on yield of spring wheat breeding, resulting primarily from greater sensitivity to EDD of the advanced spring wheat breeding lines than CK. Spring wheat typically is at floral stage during a higher temperature cycle than winter wheat during its floral stage. " in lines 251-254. Line 225: add "variety" Reply: We have removed the sentence and now it is: " Warming poses a more harmful effect on yield of spring wheat breeding, resulting primarily from greater sensitivity to EDD of the advanced spring wheat breeding lines than CK. Spring wheat typically is at floral stage during a higher temperature cycle than winter wheat during its floral stage. " in lines 251-254. Line 334-335: I would like to know about management used. How much irrigation, fertilizer, pesticides were used? Did that change every year? Because it would be another reason for changes in yield over time. Reply: The agronomic management information is not available to the published data of USDA. So, we add site-year terms in panel data model to remove the effects of non-climate factors (line 360). In our manuscript, we focus on yield changes due to climate, rather than non-climate factors. We have revised: "Unfortunately, only a small subset of sites recorded phenology data, and therefore trial-specific growing season lengths could not be used without omitting a large fraction of the data. The average plant and harvest dates in each site therefore were used based on Sacks et al. for both winter and spring wheat. A fixed time window was often set for some temperature accumulation indices in climate impact studies, rather than the growing season of each individual year. This is because the latter would result in endogeneity in an analysis of FDD, GDD and EDD on yields (e.g. warmer season may not have higher value of GDD as shorter growing season). " in lines 345-352. Despite the endogeneity problem, we can still test whether this will influence our results. We randomly selected the planting and harvest dates in different years for a certain site based on the calendar range reported in Sacks et al. And we repeat the process for 1000 times, and did the regression analysis (Fig. R2). We found that: (1) The yield impact by 1 o C warming is slightly smaller than the results in Fig. 2. This reflects the endogeneity problem, and the degree-day cannot well reflect the actual temperature difference between years. (2) The major conclusion in the main text still holds: the yield decline of MYG is smaller than CK for winter wheat, the results is contrary for spring wheat. Therefore, we believe crop calendar does not influence our major conclusion. Line 495: The link did not work when I tried it (April 2022). Please replace. Even better would be to have the data in a data repository with a permalink and DOI. Reply：We have checked the Internet links of winter wheat nurseries. The link works from my place. Please check it below: https://www.ars.usda.gov/plains-area/lincoln-ne/wheat-sorghum-and-forage-research/docs/hard-w inter-wheat-regional-nursery-program/research/ The data belongs to USDA. And is already publically available online. See the links to the data that are provided. Reviewers' Comments: Reviewer #1: Remarks to the Author: Reviewer Blind Comments to Author The authors have adequately addressed my comments and those from the other two reviewers. A main weakness of the article is the exclusion of CO2 fertilization in regression models of historic yields and subsequent yield projection into year 2100 under future climate. I have a few minor comments related to the revision of the manuscripts listed below. Specific comments L32: Results shows --> Results show L33: 3.6%/oC --> 3.6% per 1˚C warming L33-34: compared with -5.4%/˚C when variety is held constant, reflecting a superior climateresilience --> compared with -5.4% for the check variety, indicating a superior climate-resilience L35: demonstrate 7.5% yield reduction per 1˚C warming. That is more sensitive than the value of -7.1%/oC with constant variety planted, indicating an undermined climate-resilience for spring wheat. --> showed a 7.5% yield reduction per 1˚C warming, which is more sensitive than a 7.1% reduction for the check variety, indicating less climate-resilience for spring wheat. L40-41: Our study highlights that following the current wheat breeding adaptation progress is challenging to abate climate warming --> How about this: Our study highlights that the adaptation progress following the current wheat breeding strategies is challenging to abating climate warming. L59: this mechanism --> these negative effects L69-70: field observations enabling comparison between yield sensitivities --> field observations, hindering comparison in yield sensitivities L77 : a constant check --> a common check L111: Increase in FDD --> You mean decrease in FDD? L143: by 7.5% yield decline --> with 7.5% yield decline L148-166: -> These might fit better to the Discussion section. L151: was repeated but used the --> was repeated using the L165-166: is not relevant to --> How about: is insensitive to L176: is mixed in direction of changes --> is mixed in the direction of changes L213: projected to be reduce to 43.8% --> projected to be reduced to 43.8% L215: Figure 4 legend: missing HYG legend L215: Figure 4 legend: Incorrect name for different SPPs L216: Figure 4 caption: there is no need to list the six climate models L232: heading phases have shifted to an earlier date over time --> Note: This could be due to shifting of planting to earlier dates. L223-241: --> These might fit better to the Discussion section. L253: Spring wheat typically is at floral stage during a higher temperature cycle than winter wheat during its floral stage --> The floral stage of spring wheat typically coincides with a higher temperature cycle than winter wheat L256: indicating winter wheat is adapting to climate warming in a positive manner --> indicating greater climate adaptability. L257-258: Such contrasting responses result in yield benefits due to variety breeding increases in winter wheat and decreases in spring wheat --> Such contrasting responses result in greater yield benefits for winter wheat than for spring wheat from breeding progress. L261: warming is more different and challenging --> warming is more challenging L288: Northward shifting area of winter wheat --> Northward shifting of winter wheat production L292: are required to be implement now --> are required to be implement now L300-303: our study presents for the first time within a 59-year-long and multi-site dataset from breeding nurseries in which yield responses of new wheat varieties to climate were compared with long-term constant CK varieties in each season under rainfed conditions in North America --> our study presented a first comprehensive analysis, using a multi-year and multi-site dataset from breeding nurseries, on yield responses of new wheat varieties to climate under rainfed conditions in North America L305-306: offset yield benefits by switching to new varieties in the assumption that the current breeding progress would continue --> offset yield gains with the continuation of current breeding progress L308-311: Innovation on breeding technologies accompanied with alternation in traditional breeding processes preparing for more extreme climate can help to ensure the future productivity and climate resilience of wheat in a changing climate --> Integration of innovative technologies with traditional approaches in breeding for future climate can help to ensure the future productivity and climate resilience of wheat in a changing climate. Reviewer #4: Remarks to the Author: The response to the review comments is thorough and very much satisfactory. The only additional minor quibble that I have is over the assertion in the abstract that the slightly higher estimated yield reduction from the advanced breed spring wheat implies that climate resilience is "undermined" by the breeding. I'm not convinced from the analysis that this represents a statistically significant difference so I suspect it would be more correct to say that "climate resilience is not improved and may even decline" or similar language.	2022-09-30T14:02:12.932Z	2022-09-30T00:00:00.000	{ "year": 2022, "sha1": "999a53dac0d295339d18a0353f12b3c7b3a425f4", "oa_license": "CCBY", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "5be6b479a3e38cac235e214bb9061fbc0a5c2089", "s2fieldsofstudy": [ "Agricultural And Food Sciences", "Environmental Science" ], "extfieldsofstudy": [ "Medicine" ] }
102353261	pes2o/s2orc	v3-fos-license	Neural Network-Based Formula for the Buckling Load Prediction of I-Section Cellular Steel Beams Cellular beams are an attractive option for the steel construction industry due to their versatility in terms of strength, size, and weight. Further benefits are the integration of services thereby reducing ceiling-to-floor depth (thus, building’s height), which has a great economic impact. Moreover, the complex localised and global failures characterizing those members have led several scientists to focus their research on the development of more efficient design guidelines. This paper aims to propose an artificial neural network (ANN)-based formula to estimate the critical elastic buckling load of simply supported cellular beams under uniformly distributed vertical loads. The 3645-point dataset used in ANN design was obtained from an extensive parametric finite element analysis performed in ABAQUS. The independent variables adopted as ANN inputs are the following: beam’s length, opening diameter, web-post width, cross-section height, web thickness, flange width, flange thickness, and the distance between the last opening edge and the end support. The proposed model shows a strong potential as an effective design tool. The maximum and average relative errors among the 3645 data points were found to be 3.7% and 0.4%, respectively, whereas the average computing time per data point is smaller than a millisecond for any current personal computer. Introduction The use of cellular beams (i.e., perforated beams with circular web openings) in the construction sector has significantly increased over the past decade on account of the distinct and discreet advantages they offer. Cellular beams are applicable for long span structures, where integration of services such as ventilation ducts and lightings systems within the beam is attained, but also for short spans, where spatial interference among concentrated mechanical devices and structural elements may require a compromised solution. Cellular beams allow reducing the height of buildings to fit a required number of floors, otherwise fitting more floors in a given height limit having a significant economic impact to the whole structure's budget. Furthermore, cellular beams offer practical advantages such as the possibility of (i) fixing the ceilings directly to the beams' lower flanges instead of requiring additional suspension elements, and (ii) allowing future equipment addition or replacement within the existent void holes. In fact, with the wider adoption of Building Information Modelling (BIM), the knowledge of those expansion possibilities is becoming a valuable asset for the building management. Long span and lightweight structures also benefit from flexible designs with the fewer number of columns and foundations, and thus from the reduced construction time (Tsavdaridis 2010). The increase in beam depth due to the castellation process (i.e., profile cutting manufacturing) also provides greater flexural stiffness having a final section with larger section modulus (Morkhade and Gupta 2015). However, the presence of web openings significantly (ASCE 23-97, SCI-P100, SCI-P355) and produce the lowest errors. In 2011, Lawson and Hicks published the SCI-P355 (Lawson and Hicks 2011) design guidelines, an update to SCI-P068 (Lawson 1987) and SCI-P100 (Ward 1990) which proposed that the Vierendeel bending resistance is dependent on the classification of the web of the T-beams. This approach produced acceptable approximations for openings of specific dimensions where the best results were found with an error of 25-30%. It is worth noting that Chung et al. (2003), Verweij (2010) and Morkhade and Gupta (2015) have reported that the current guidelines, specifically SCI-P100 (Ward 1990) and SCI-P355 (Lawson and Hicks 2011), are inadequate, complicated and conservative when it comes to the design of perforated steel beams. Artificial Neural Networks (ANN) have become a popular method to predict the response of structures. presented a study relating the use of ANN in the evaluation of the load carrying capacity of the web-post of castellated steam beams based on 140 FE models. The computational technique generated predictions with great accuracy when compared to other methods. Sharifi and Tohidi (2014) also illustrated the application of ANN to accurately estimate the elastic buckling load capacity of steel bridge girders that have rectangular openings at the bottom zone in the web. This is considered as the worse possible location to place an opening to resist lateral torsional buckling. The ANN formula was derived from 21 FE models which managed to accurately predict the elastic buckling load. In 2014, . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 © 2018 by Abambres et al. (CC BY 4.0) 6 Tohidi and Sharifi demonstrated the versatility of ANN by studying the buckling capacity of steel beams with rectangular web openings that has experienced corrosion in the web. In addition, Tohidi and Sharifi (2015) developed an ANN model to estimate the bearing capacity of steel girders with corrosion at the bearing region. The ANN empirical formulas obtained were reported to be accurate in predicting the residual capacity of deteriorated steel beams. The current study was motivated by the lack of rational (simple, efficient and accurate) design procedures relating to the buckling response of cellular beams. This paper proposes an ANN-based formula to estimate the critical elastic buckling load of simply supported cellular beams under uniformly distributed vertical loads, as function of eight independent geometrical parameters. This research is the first step of an ongoing investigation that aims to propose a novel and simple analytical design method to accurately compute the inelastic resistance of cellular steel beams. Α FE-based dataset comprising 3645 points was generated for this study, in order to allow the ANN model to have a significant generalization ability and be considered as a powerful tool for structural engineers and scientists to (i) estimate the elastic buckling load of cellular steel beams, and (ii) efficiently perform sensitivity analyses to further assess the behaviour of those members. . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 © 2018 by Abambres et al. (CC BY 4.0) 7 FE Modelling Three-dimensional FE models were developed using ABAQUS (Dassault Systèmes Simulia Corp, 2017), which were then parametrised to generated 3645 simulations. Typical values for the modulus of elasticity and Poisson's ratio were adopted (E = 210 GPa, ν=0.3). All models are simply supported where one end allows in-plane rotations but not translations and the other admits translations along the beam axis, beyond in-plane rotations. End twisting rotations were prevented by restraining both the top and bottom flange tips against out-of-plane displacements at the supports. A unitary load was applied to the top flange as a uniformly distributed pressure (then converted to a line load for ANN simulation purposessee Tab. 1). The FE mesh adopted was quad-dominated using shell elements of type SR8, which was tested against experimental work conducted by Tsavdaridis andD'Mello (2011), andSurtees andLiu (1995), providing accurate and reliable results (Rajana 2018). The mesh sizes recommended by El-Sawy et al. (2014) for web and flanges were adopted. Fig. 1(a) illustrates the various parameters considered in the parametric analysis, whereas Fig. 1(b) illustrates one application of these type of structural members. . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 Parametric Analysis The parametric models were submitted to the ABAQUS Lanczos Eigensolver using Python scripts. Tab. 1 presents the possible values taken for each independent (parametric) variable (see Fig. 1(a)) considered in the FEA. The 'first' web opening was placed at the centre of the beam whereas the remaining ones were offset from the former until (for a fixed beam's length, opening diameter, and web-post width) no more circular openings could fit within member's . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 © 2018 by Abambres et Introduction Machine learning, one of the six disciplines of Artificial Intelligence (AI) without which the task of having machines acting humanly could not be accomplished, allows us to 'teach' computers how to perform tasks by providing examples of how they should be done (Hertzmann and Fleet 2012). When there is abundant data (also called examples or patterns) . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 © 2018 by Abambres et al. (CC BY 4.0) 10 explaining a certain phenomenon, but its theory richness is poor, machine learning can be a perfect tool. The world is quietly being reshaped by machine learning, being the Artificial Neural Network (also referred in this manuscript as ANN or neural net) its (i) oldest (McCulloch and Pitts 1943) and (ii) most powerful (Hern 2016) technique. ANNs also lead the number of practical applications, virtually covering any field of knowledge Irwin 2011, Prieto et. al 2016). In its most general form, an ANN is a mathematical model designed to perform a particular task, based in the way the human brain processes information, i.e. with the help of its processing units (the neurons). ANNs have been employed to perform several types of real-world basic tasks. Concerning functional approximation, ANN-based solutions are frequently more accurate than those provided by traditional approaches, such as multi-variate nonlinear regression, besides not requiring a good knowledge of the function shape being modelled (Flood 2008). The general ANN structure consists of several nodes disposed in L vertical layers (input layer, hidden layers, and output layer) and connected between them, as depicted in Fig. 2. Associated to each node in layers 2 to L, also called neuron, is a linear or nonlinear transfer (also called activation) function, which receives the so-called net input and transmits an output (see Fig. 5). All ANNs implemented in this work are called feedforward, since data presented in the input layer flows in the forward direction only, i.e. every node only connects to nodes belonging to layers located at the right-hand-side of its layer, as shown in Fig. 2. ANN's computing power makes them suitable to efficiently solve small to large-scale complex . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 © 2018 by Abambres et al. (CC BY 4.0) 11 problems, which can be attributed to their (i) massively parallel distributed structure and (ii) ability to learn and generalize, i.e., produce reasonably accurate outputs for inputs not used during the learning (also called training) phase. Learning Each connection between 2 nodes is associated to a synaptic weight (real value), which, together with each neuron's bias (also a real value), are the most common types of neural net unknown parameters that will be determined through learning. Learning is nothing else than determining network unknown parameters through some algorithm in order to minimize network's performance measure, typically a function of the difference between predicted and target (desired) outputs. When ANN learning has an iterative nature, it consists of three phases: should not be applied outside the input variable ranges used for network training. Since there performance results for each designed net, thus allowing the selection of the best ANN according to a certain criterion. The best network in each parametric SA is the one exhibiting the smallest average relative error (called performance) for all learning data. It is worth highlighting that, in this manuscript, whenever a vector is added to a matrix, it means the former is to be added to all columns of the latter (valid in MATLAB). Dimensional Analysis (feature 2) The most widely used form of dimensional analysis is the Buckingham's π-theorem, which was implemented in this work as described in Bhaskar and Nigam (1990). Input Dimensionality Reduction (feature 3) When designing any ANN, it is crucial for its accuracy that the input variables are independent and relevant to the problem , Kasun et al. 2016). There are two types of dimensionality reduction, namely (i) feature selection (a subset of the original set of input variables is used), and (ii) feature extraction (transformation of initial variables into a smaller set). In this work, dimensionality reduction is never performed when the number of input variables is less than six. The implemented methods are described next. where (i) Var(X) and Cov(X, Y) are the variance of X and covariance of X and Y, respectively, and (ii) ̅ and ̅ are the mean values of each variable. In this work, cases where \| \| ≥ 0.99 indicate that one of the variables in the pair must be removed from the ANN modelling. The one to be removed is the one appearing less in the remaining pairs ( , ) where \| \| ≥ 0.99. Once a variable is selected for removal, all pairs ( , ) involving it must be disregarded in the subsequent steps for variable removal. Auto-Encoder This feature extraction technique uses itself a 3-layer feedforward ANN called auto-encoder (AE). After training, the hidden layer output (y2p) for the presentation of each problem's input pattern (y1p) is a compressed vector (Q2 x 1) that can be used to replace the original input layer by a (much) smaller one, thus reducing the size of the ANN model. In this work, Q2=round(Q1/2) was Concerning the learning algorithm used for all AEs, no L2 weight regularization was employed, which was the only default specification not adopted in 'trainAutoencoder(…)'. Orthogonal and Sparse Random Projections This is another feature extraction technique aiming to reduce the dimension of input data Y1 (Q1 x P) while retaining the Euclidean distance between data points in the new feature space. This is attained by projecting all data along the (i) orthogonal or (ii) sparse random matrix A (Q1 x Q2, Q2 < Q1), as described by Kasun et al. (2016) Training, Validation and Testing Datasets (feature 4) Four distributions of data (methods) were implemented, namely pt-pv-ptt = {80-10-10, 70- 3) In order to select the validation patterns, randomly select pv / (pv + ptt) of those patterns not belonging to the previously defined training dataset. The remainder defines the testing dataset. It might happen that the actual distribution pt-pv-ptt is not equal to the one imposed a priori (before step 1), which is due to the minimum required training patterns specified in step 2. Input Normalization (feature 5) The progress of training can be impaired if training data defines a region that is relatively narrow in some dimensions and elongated in others, which can be alleviated by normalizing each input variable across all data patterns. The implemented techniques are the following: Linear Max Abs Lachtermacher and Fuller (1995) proposed a simple normalization technique given by . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 Nonlinear Proposed by Pu and Mesbahi (2006), although in the context of output normalization, the only nonlinear normalization method implemented for input data reads The Hyperbolic Tangent function is also of sigmoid type, being defined as . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 Normalization can also be applied to the output variables so that, for instance, the amplitude of the solution surface at each variable is the same. Otherwise, training may tend to focus (at least in the earlier stages) on the solution surface with the greatest amplitude (Flood and Kartam 1994a). Normalization ranges not including the zero value might be a useful alternative since convergence issues may arise due to the presence of many small (close to zero) target values (Mukherjee et al. 1996). Four normalization methods were implemented. The first three follow eq. The fourth normalization method implemented is the one described by eq. (6). Multi-Layer Perceptron Network (MLPN) This is a feedforward ANN exhibiting at least one hidden layer. Fig. 2 depicts a 3-2-1 MLPN (3 input nodes, 2 hidden neurons and 1 output neuron), where units in each layer link only to some nodes located ahead. At this moment, it is appropriate to define the concept of partially-(PC) and fully-connected (FC) ANNs. In this work a FC feedforward network is characterized by having each node connected to every node in a different layer placed forwardany other type of network is said to be PC (e.g., the one in Fig. 2). According to Wilamowski (2009), PC MLPNs are less powerful than MLPN where connections across layers are allowed, which usually lead to smaller networks (less neurons). where φl is the transfer function used for all neurons in layer l. Radial-Basis Function Network (RBFN) Although having similar topologies, RBFN and MLPN behave very differently due to distinct hidden neuron modelsunlike the MLPN, RBFN have hidden neurons behaving differently than output neurons. According to Xie et al. (2011), RBFN (i) are specially Lastly, according to the implementation carried out for initialization purposes (described in 3.3.12), (i) RBF center vectors per hidden layer (one per hidden neuron) are initialized as integrated in a matrix (termed RBF center matrix) having the same size of a weight matrix linking the previous layer to that specific hidden layer, and (ii) RBF widths (one per hidden neuron) are initialized as integrated in a vector (called RBF width vector) with the same size of a hypothetic bias vector. Connectivity (feature 10) For this ANN feature, three methods were implemented, namely (i) adjacent layersonly connections between adjacent layers are made possible, (ii) adjacent layers + input-outputonly connections between (ii1) adjacent and (ii2) input and output layers are allowed, and (iii) fully-connected (all possible feedforward connections). . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 © 2018 by Abambres et al. (CC BY 4.0) (9), defined in 3.3.6, the ones defined next were also implemented as hidden transfer functions. During software validation it was observed that some hidden node outputs could be infinite or NaN (not-a-number in MATLABe.g., 0/0=Inf/Inf=NaN), due to numerical issues concerning some hidden transfer functions and/or their calculated input. In those cases, it was decided to convert infinite to unitary values and NaNs to zero (the only exception was the bipolar sigmoid function, where NaNs were converted to -1). Other implemented trick was to convert possible Gaussian function's NaN inputs to zero. Identity-Logistic In Gunaratnam and Gero (1994), issues associated with flat spots at the extremes of a sigmoid function were eliminated by adding a linear function to the latter, reading where \|\| … \|\| denotes the Euclidean distance in all functions. Parameter Initialization (feature 12) The initialization of (i) weight matrices (Qa x Qb, being Qa and Qb node numbers in layers a and b being connected, respectively), (ii) bias vectors (Qb x 1), (iii) RBF center matrices (Qc-1 x Qc, being c the hidden layer that matrix refers to), and (iv) RBF width vectors (Qc x 1), are independent and in most cases randomly generated. For each ANN design carried out in the context of each parametric analysis combo, and whenever the parameter initialization method is not the 'Mini-Batch SVD', ten distinct simulations varying (due to their random nature) initialization values are carried out, in order to find the best solution. The implemented initialization methods are described next. . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 Rand [-lim, lim] This function is based on the proposal in Waszczyszyn (1999), and generates random numbers with uniform distribution in [-lim, lim], being lim layer-dependent and defined by where a and b refer to the initial and final layers integrating the matrix being initialized, and L is the total number of layers in the network. In the case of a bias or RBF width vector, lim is always taken as 0.5. . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 © 2018 by Abambres et al. (CC BY 4.0) SVD Although Deng et al. (2016) proposed this method for a 3-layer network, it was implemented in this work regardless the number of hidden layers. Mini-Batch SVD Based on Deng et al. (2016), this scheme is an alternative version of the former SVD. Now, training data is split into min{Qb, Pt} chunks (or subsets) of equal size Pti = max{floor(Pt / Qb), 1}floor rounds the argument to the previous integer (whenever it is decimal) or yields the argument itself, being each chunk aimed to derive Qbi = 1 hidden node. Learning Algorithm (feature 13) The most popular learning algorithm is called error back-propagation (BP), a first-order gradient method. Second-order gradient methods are known to have higher training speed and accuracy (Wilamowski 2011). The most employed is called Levenberg-Marquardt (LM). All these traditional schemes were implemented using MATLAB toolbox (The Mathworks, Inc 2017). Concerning the LM scheme -'trainlm' in MATLAB, the only learning parameter set different than its default value was the abovementioned (ii). Extreme Learning Machine (ELM, mb ELM, I-ELM, CI-ELM) Besides these traditional learning schemes, iterative and time-consuming by nature, four Performance Improvement (feature 14) A simple and recursive approach aiming to improve ANN accuracy is called Neural Network Composite (NNC), as described in Beyer et al. (2006). In this work, a maximum of 10 extra ANNs were added to the original one, until maximum error was not improved between successive NNC solutions. Later in this manuscript, a solution given by a single neural net might be denoted as ANN, whereas the other possible solution is called NNC. . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 Wilson and Martinez (2003) suggested that if one wants to use mini-batch training with the same stability as online training, a rough estimate of the suitable learning rate to be used in learning algorithms such as the BP, is ηonline /√ , where cs is the chunk size and ηonline is the online learning ratetheir proposal was adopted in this work. Based on the proposal of Liang et al. (2006), the constant chunk size (cs) adopted for all chunks in mini-batch mode reads cs = min{mean(hn) + 50, Pt}, being hn a vector storing the number of hidden nodes in each hidden layer in the beginning of training, and mean(hn) the average of all values in hn. Network Performance Assessment Several types of results were computed to assess network outputs, namely (i) maximum error, (ii) % errors greater than 3%, and (iii) performance, which are defined next. All where (i) dqp is the q th desired (or target) output when pattern p within iteration i (p=1,…, Pi) is presented to the network, and (ii) yqLp is net's q th output for the same data pattern. Moreover, denominator in eq. (25) is replaced by 1 whenever \|dqp\| < 0.05dqp in the nominator keeps its real value. This exception to eq. (25) aims to reduce the apparent negative effect of large relative errors associated to target values close to zero. Even so, this trick may still lead to (relatively) large solution errors while groundbreaking results are depicted as regression plots (target vs. predicted outputs). Maximum Error This variable measures the maximum relative error, as defined by eq. (25), among all output variables and learning patterns. . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 Percentage of Errors > 3% This variable measures the percentage of relative errors, as defined by eq. (25), among all output variables and learning patterns, that are greater than 3%. Performance In functional approximation problems, network performance is defined as the average relative error, as defined in eq. (25), among all output variables and data patterns being evaluated (e.g., training, all data). Software Validation Several benchmark datasets/functions were used to validate the developed software, involving low-to high-dimensional problems and small to large volumes of data. Due to paper length limit, validation results are not presented herein but they were made public by Researcher (2018b). Parametric Analysis Results Aiming to reduce the computing time by cutting in the number of combos to be runnote that all features combined lead to hundreds of millions of combos, the whole parametric simulation was divided into nine parametric SAs, where in each one feature 7 only takes a single value. This measure aims to make the performance ranking of all combos within each 'small' analysis more . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 and SA 2 involved the ELM-based counterpart, (ii) the 3 rd -7 th SAs combined all possible methods from features 3, 4, 6 and 7, and concerning all other features, adopted the methods integrating the best combination from the aforementioned first SA, (iii) the 8 th SA combined all possible methods from features 11, 12 and 14, and concerning all other features, adopted the methods integrating the best combination (results compared after postprocessing) among the previous five sub-analyses, and lastly (iv) the 9 th SA combined all possible methods from features 9, 10 and 15, and concerning all other features, adopted the methods integrating the best combination from the previous analysis. Summing up the ANN feature combinations for all parametric SAs, a total of 475 combos were run for this work. ANN feature methods used in the best combo from each of the abovementioned nine parametric sub-analyses, are specified in Tab. 5 (the numbers represent the method number as in Tabs 2-4). Proposed ANN-Based Model The proposed model is the one, among the best ones from all parametric SAs, exhibiting the lowest maximum error (SA 9 with 5 layers and a distribution of nodes/layer of 8-11-11-11-1. Concerning connectivity, the network is fully-connected, and the hidden and output transfer functions are all Logistic (eq. (7)) and Identity (eq. (10)), respectively. The network was trained using the LM algorithm (1500 epochs). After design, the average network computing time concerning the presentation of a single example (including data pre/postprocessing) is 6.93E-05 s - Fig. 7 It is worth recalling that, in this manuscript, whenever a vector is added to a matrix, it means the former is to be added to all columns of the latter (valid in MATLAB). 2.6 0.3 0.0 32 8.34E-05 7 9.5 0.9 3.9 29 6.93E-05 8 7.0 0.7 1.9 29 6.80E-05 9 -----(b) Input Data Preprocessing For future use of the proposed ANN to simulate new data Y1,sim (8 x Psim matrix) concerning Psim patterns, the same data preprocessing (if any) performed before training must be applied to the input dataset. That preprocessing is defined by the methods used for ANN features 2, 3 and 5 (respectively 2, 6 and 1see Tab. 2), which should be applied after all (eventual) qualitative . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 © 2018 by Abambres et al. (CC BY 4.0) 42 variables in the input dataset are converted to numerical (using feature 1's method). Next, the necessary preprocessing to be applied to Y1,sim, concerning features 2, 3 and 5, is fully described. Dimensional Analysis and Dimensionality Reduction Since dimensional analysis (d.a.) and dimensionality reduction (d.r.) were not carried out, one has     . Neural Network-based formula for the buckling load prediction of I-section cellular steel beams, hal-02074835 ANN-Based Analytical Model Once determined the preprocessed input dataset {Y1,sim}n after (8 x Psim matrix), the next step is to present it to the proposed ANN to obtain the predicted output dataset {Y5,sim}n after (1 x Psim vector), which will be given in the same preprocessed format of the target dataset used in learning. In order to convert the predicted outputs to their 'original format' (i.e., without any transformation due to normalization or dimensional analysisthe only transformation visible will be the (eventual) qualitative variables written in their numeric representation), some postprocessing is needed, as described in detail in 3.7.3. Next, the mathematical representation of the proposed ANN is given, so that any user can implement it to determine {Y5,sim}n after , thus eliminating all rumors that ANNs are 'black boxes'. . (31) Arrays Wj-s and bs are stored online in Developer (2018b), aiming to avoid an overlong article and ease model's implementation by any interested reader. Output Data Postprocessing In order to transform the output dataset obtained by the proposed ANN, {Y5,sim}n after (1 x Psim vector), to its original format (Y5,sim), i.e. without the effects of dimensional analysis and/or output normalization (possibly) taken in target dataset preprocessing prior training, the postprocessing addressed next must be performed. Once obtained {Y5,sim}n after , the following relations hold for its relation to its non-normalized ({ 5, } . . ) and original (Y5,sim) formats (just after the dimensional analysis stage, and free of any pre-processing effects, respectively), reading	2019-04-09T13:01:48.143Z	2018-12-01T00:00:00.000	{ "year": 2018, "sha1": "384da6db49a6db2941c5032d52f7137d2a7ebdde", "oa_license": "CCBY", "oa_url": "https://hal.archives-ouvertes.fr/hal-02074835/file/Abambres%20et%20al%202018.pdf", "oa_status": "GREEN", "pdf_src": "ElsevierPush", "pdf_hash": "40b7feab246cb59774324dcb1f8bdcfe25f21645", "s2fieldsofstudy": [ "Engineering" ], "extfieldsofstudy": [ "Computer Science" ] }
251320651	pes2o/s2orc	v3-fos-license	RKKY to Kondo crossover in Helical Edge of a Topological Insulator Two spatially separated magnetic impurities coupled to itinerant electrons give rise to a dynamically generated exchange (RKKY) inter-impurity interaction that competes with the individual Kondo screening of the impurities. It has been recently shown by Yevtushenko and Yudson (2018), that the RKKY interaction and the RKKY vs. Kondo competition become nontrivial on helical edges of two-dimensional topological insulators where there is lock-in relation between the electron spin and its direction of motion. Kondo screening always takes over and dominates at large inter-impurity distances and it can also dominate all the way to short distances if the Kondo coupling is sufficiently large and anisotropic. In the present paper, we study the Kondo-RKKY competition in detail on a qualitative and quantitative level. For this we employ the numerically exact numerical renormalization group (NRG) for a broad parameter scan of two Kondo coupled impurities vs. magnetic anisotropy, impurity distance and temperature, and comment on the role of finite bandwidth. We give a pedagogical introduction on the the setup of the two-impurity setting within the NRG in the helical context. Overall we establish a plain crossover from RKKY to Kondo with increasing impurity distance which permits an intuitive physical picture by simply comparing length scales set by the Kondo screening cloud, the thermal length scale vs. the impurity distance. I. INTRODUCTION Physics of magnetic impurities (MIs) coupled to helical electrons on one-dimensional (1D) edges of two dimensional (2D) time-reversal invariant topological insulators (TIs) attracted the attention of researchers soon after the experimental discovery of the TIs [2][3][4][5].This interest resulted from a search of possible backscattering mechanisms which could make the virtually protected helical conductance sub-ballistic in relatively long samples [6][7][8][9].Since the helical electrons possess a lock-in relation between the spin projection on the quantization axis and the direction of propagation (the so-called chirality), their backscattering is expected to involve some nontrivial spin processes, e.g. the spin flip.The MI can provide such an inelastic backscattering of the individual helical electrons.However, the helical conductance can be suppressed only if the spin conservation on the edge is violated, see papers [10,11] and references therein.If the edge is spinconserving and the MI does not break the spin U(1) symmetry, it backscatters the helical electrons but cannot influence the dc conductance [12].The anisotropic MI is able to suppress the helical conductance only if it breaks the spin conservation and is not Kondo screened [13][14][15].The latter requires either the temperature being larger than the Kondo temperature, T > T K , or a large value of the MI spin, S > 1/2.This points out the importance of understanding the Kondo effect in TIs which is substantially different from that in usual (non-helical) 1D wires in the presence of the electron-electron interaction or the magnetic anisotropy of the XXZ type [16]. Here as well as throughout this paper, T K ≡ T (1) K represents * weichselbaum@bnl.gov the Kondo scale of a single, possibly anisotropic spin-half impurity coupled to a helical edge [16].Its value is identical to the plain non-helical Kondo model given that that helicity in the non-interacting bath is irrelevant from the point of view of a single impurity. The Kondo effect can be suppressed by the indirect exchange MI interaction, the Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction [17], if the helical edge is coupled to a dense array of the MIs.According to the simple picture of the Doniach criterion [18], the "winner of the RKKY-Kondo competition" can be found by comparing T K with the characteristic RKKY energy scale E R .The latter means the energy gap which opens after the RKKY correlations lift a degeneracy in the energy of the uncorrelated MIs.If T → 0 but E R ≫ T K , the RKKY correlations overwhelm the Kondo screening which may lead to many nontrivial effects, including Anderson localization of the helical electrons caused by the random magnetic anisotropy [19,20], and magnetically correlated phases [21][22][23][24].The RKKY-induced magnetic order in helical 1D systems is nontrivial and qualitatively different from that in their non-helical counterparts, cf.Refs.[25,26] and references therein. Since T K does not depend on the MI density while E R typically decays with increasing the inter-impurity distance, R -see Eq. (20) below, one can surmise that there is a characteristic distance defined by the equality T K ≃ E R (x c ), which separates the RKKY-and Kondo-dominated phases, R < x c and R > x c respectively.This conclusion is inspired by a misleading analogy with the physics of non-helical wires [27][28][29][30][31].One of us (in collaboration with V. I. Yudson [1]) have recently considered two MIs coupled to the helical edge and shown that, if either the electron-electron interaction or the magnetic XXY anisotropy or both are strong, x c shrinks and the Kondo effect overwhelms the RKKY in-arXiv:2208.02275v3[cond-mat.str-el] 1 Sep 2023 teraction over all macroscopic inter-impurity distances.This unexpected conclusion has been drawn based on phenomenological arguments and on the analytical consideration of limiting cases.This theory reveals many quantitative features but is far from being complete.In particular, it cannot predict whether the above mentioned phases with finite and vanishing x c are separated by a crossover or by a phase transition and whether the MIs remain somehow correlated even in the Kondo-dominated phase. In the present paper, we expand and complete the theory of Ref. [1].We present an analytical theory of the RKKY correlation between the two MIs coupled to the helical electrons but, as the main working tool, we have chosen the numerical renormalization group (NRG; [32][33][34]).This powerful and well-established method has allowed us to answer the aforementioned open questions.In particular we will show that 1) the different phases are separated by the crossover, and 2) the MIs are uncorrelated, i.e. independently screened, in the Kondo dominated phase. The paper is organized as follows: We introduce the model in Sec.II, followed by analytical considerations in Sec.III.The remainder of the paper then is dedicated to a detailed analysis and discussion of the model based on the NRG in Sec.IV, concluded by summary and outlook.In appendix App.A, we give a detailed pedagogical derivation of how the helical 2-impurity system is setup and mapped into the standard NRG machinery.In particular, this highlights the possibility to map the system onto a Wilson ladder, with complex coefficients only within the coupling to the impurity.Furthermore, we included in the appendix a brief reminder on the poor-mans scaling of the anisotropic Kondo model, as well as a plain second order perturbative derivation of the RKKY effective Hamiltonian and RKKY energy that is complementary to Sec.III. 1. Helical edge mode of the one-dimensional edge in a 2D topological insulator, e.g., as it occurs in the Kane-Mele model [35].Focusing on low energies, the dispersion vs. momentum k is given by ε kσ = σvk with spin σ ∈ {↑, ↓} ≡ {+1, −1} indicated by the blue (red) line, respectively, where v denotes Fermi velocity.We assume a finite half-bandwidth D, throughout.This gives rise to an effective Brillouin zone with lattice spacing a [cf.Eq. ( 2)]. II. THE MODEL A. Hamiltonian of the helical edge We study spins coupled to a helical edge mode in a 2D topological insulator.This may be approached in various ways.The edge mode can be simulated (i) by fully modeling an underlying 2D lattice model in real space, such as the Kane-Mele model with the Dresselhaus spin-orbit interaction [35].This naturally introduces a cutoff in terms of bandwidth of the helical edge mode which is located inside the gap of the continuum of extended bulk states.Together with interacting correlated impurities, this may be simulated numerically, for example, using the density matrix renormalization group (DMRG, [36,37]).There the non-interacting 2D lattice without the impurities can be conveniently mapped to an effective 1D impurity setting via Lanczos tridiagonalization [38].However, this approach bears significant overhead in terms of the precise choice of the underlying 2D lattice model and its parameters, the details of which are considered irrelevant for the low-energy physics.Conversely, (ii) the model can be considerably simplified by focusing on a single pure effective 1D helical edge described in energy-momentum space at low energies as depicted in Fig. 1 and described by Eq. ( 1) below.The latter representation of the ballistic non-interacting edge modes may be (iii) exactly transformed into a 1D real-space lattice realization which, however, involves long-range hoppings [cf.Sec.A 2 and in particular Eq.(A12) for more on this].Working in energy-momentum space, instead, is appealing from an analytical point of view [1], but is also perfectly well-suited for the numerical treatment via the NRG.With the additional goal to scan many orders of energy scales with Kondo physics in mind, the energy-momentum representation is preferred over real-space lattice descriptions.Therefore we follow approach (ii) throughout this work. The standard model Hamiltonian for a single helical edge mode can be written in discrete form in momentum space as follows [cf.Fig. 1], where ĉkσ are the annihilation operators of the helical fermions with spin σ ∈ {↑, ↓} ≡ {+1, −1}, and τ α with α ∈ x, y, z the Pauli matrices.The fermions are described by a linearized dispersion relation with the Fermi velocity v. Below we consider a finite half-bandwidth D (UV cutoff) such that ε kσ ∈ [−D, D].This is required in the numerical context yet and also for the sake of regularization.We assume that the TI edge is oriented along the x-axis, with xy being the TI plane, and the z-direction is the quantization axis for the spins of the helical edge modes.This convention corresponds to the experimentally relevant situation where the quantization axis is often fixed being perpendicular to the TI plane [6][7][8][9].The helical system in Eq. ( 1) respects time-reversal symmetry (TRS), in that ε kσ = ε −k,−σ .The crossing of the spin selective dispersions, i.e., the Dirac point in Fig. 1 is chosen for simplicity at k 0 = 0 without restricting the case.The precise choice of k 0 is irrelevant for our purposes, as it can be absorbed into the definition of the basis states, and hence can be gauged away.[39] By assuming a finite half-bandwidth D and having translational invariance by working in momentum space, this directly implies an effective Brillouin zone (BZ) with momentum range \|k\| ≤ k max , see Fig. 1, with a discontinuous dispersion across the Brillouin zone (BZ) boundary.Conversely, this defines an effective lattice constant, a = π/k max , via the one-particle dispersion, having \|ε( π a )\| = D, i.e., a ≡ πv D . (2) Below, we use the local density of states ρ 0 = aρ 1D as experienced by an impurity where ρ 1D = 1/2πv is the constant one-particle density of states of a 1D system with linear dispersion relation [40] (thus units are [ρ 0 ] = 1/energy, whereas [ρ 1D ] = 1/(energy • distance)).Using Eq. ( 2), the local density of states becomes ρ 0 = 1 2D , which is consistent with standard NRG conventions.For more on the effects of finite bandwidth and an effective 1D lattice defined by Eq. ( 2), see Sec.A 2 [cf.Eq. (A11)]). In numerical simulations, furthermore, adhering to standard NRG conventions, we choose the unit of energy D := 1.By also setting the unit of distance a := 1, this fixes the velocity to v = 1/π.We also set ℏ = k B = 1. The last expression in Eq. ( 1) provides a more compact notation using the spinor ĉk ≡ (ĉ k↑ ; ĉk↓ ), where the semicolon denotes a column vector.This explicitly shows that the SU(2) spin symmetry is broken in the helical setting.It reduces to the abelian U(1) symmetry with preserved component of the total spin S tot z [this is in contrast to a chiral system where both spins move in the same direction [40,41], which thus preserves SU(2) spin symmetry].In the helical case, we thus only explore the combination of abelian symmetries U(1) charge ⊗ U(1) spin [see Sec.A 5 for further comments on symmetries]. The Hamiltonian (1) can be written in real-space in the standard continuous form where Ψ ≡ ( ΨR ; ΨL ) is the spinor constructed from the slow helical fields of the right and left moving electrons ΨR,L .For the free electrons in the helical edge mode, i.e., the bath to which the spin impurities are coupled, spins up and down and, respectively, chirality (the direction of propagation) of the right/left movers can be denoted in more general fashion by the index in particular, ΨR,L ≡ Ψ↑,↓ .When used as variable in equations below, the particular meaning of this index is always clear in context.The Hamiltonian Eq. ( 3) may be defined on a finite-length system with periodic boundary conditions (BCs) and hence discrete momenta, or in the thermodynamic limit with a continuous energy-momentum space.Sec.II A represents the simplest effective model that describes a single helical edge, for example, in HgTe/CdTe quantum-well heterostructures that possess axial and inversion symmetry around the growth axis. B. Coupling between the helical fermions and MIs Let us introduce two MI spins Ŝα η separated at distance x and located symmetrically around the origin at positions for left and right impurity, respectively [to be differentiated from right / left movers denoted by σ ∈ {R, L} in Eq. ( 4)]. By working in energy-momentum space, the bath operators at the location of the impurity are obtained via Fourier transform [e.g., cf.App.A].With this, one can focus both, analytically and numerically, on the two impurities being located along a single edge without having to worry about periodic boundary conditions.This is possible, when the actual (2D) sample is always considered much larger than the impurity distance x. Thus without restricting the case, one is free to think of the two impurities as being symmetrically located at ±x/2 along a straight edge around some arbitrary but fixed origin.The exchange interaction between the helical electrons and these two MIs is described by the Hamiltonian Here j 0 ≡ ρ 0 J and j z ≡ ρ 0 J z are the constant dimensionless exchange couplings, such that, for example, 2πv j 0 = aJ with J the coupling strength of the impurity in units of energy and [aσ α η ] = 1 dimensionless as typically used within the NRG, having with τ ± ≡ 1 2 (τ x ± iτ y ).The exchange interaction (6) may be anisotropic, having J ̸ = J z , while it always conserves the z-projection of the total (electron and MIs) spin. In the next section, we adhere to the standard notations of the literature devoted to the analytical study of Kondo impurities coupled to helical electrons, where the Kondo coupling is measured in units of the Fermi velocity, J = (2πv)j, and similarly for Jz . III. ANALYTICAL THEORY OF THE RKKY REGIME Let us briefly review the RKKY theory for the helical edge mode coupled to two MIs [1] using field theoretical machinery. A. Non-interacting fermions To describe degrees of freedom of the MIs, one should approximately integrate out the helical fermionic edge modes.A natural way to do this is to exploit the formalism of functional integrals.As a result, one arrives at an effective action for the impurity spins: Here S 0 and S int are the action of the free electron system, i.e., the bath, and of the electron-impurity interaction (6), respectively; Z 0 is the statistical sum of the electron system without spin impurities.The functional integration is performed over fermionic (Grassmann) variables.The spin degrees of freedom in the action are described by one of the known approaches (e.g.coherent spin representation or a set of Majorana Grassmann variables, etc., see the book [42]).A particular choice of the spin variable is not important for our current purposes. Let us start from the simplest case of the non-interacting fermions.The electron action S 0 in the Matsubara representation reads with σ = ±1 as in Eq. ( 4).The Matsubara Green's functions of the helical electrons in the momentum-frequency and space-frequency representation are given by The combined action reads where Calculating the Gaussian integral over the Grassmann variables, we obtain the contribution to the spin action, This expression is formally exact and, being properly regularized, describes all effects of the electron coupling to the spin impurities.Following the standard RKKY scheme, let us now focus on the weak coupling regime by restricting ourselves to terms up to second order in J in the action.This yields where X j ≡ (τ j , x j ).As argued below, other second order combinations do not contribute.After Fourier transform to the Matsubara frequencies, the first term in (15) takes the form with x ≡ x 2 − x 1 > 0 being the inter-impurity distance.The most interesting is the low temperature regime, x ≪ L T ≡ v/T , where the summation over frequencies in (16b) can be replaced by the integration over dω 2πT , resulting in Expressions similar to Eqs. ( 16) are governed also by the second term in (15).Combining all terms together, we obtain We are interested in slow motion of the MI spins with characteristic frequencies being much smaller than the inverse timeof-flight of the electron between the MIs, \|Ω n \| x v ≪ 1.In this case, e − x\|Ωn \| v ≃ 1 and, returning back to the imaginary time, we arrive at the expression This is the action of a system described by an effective RKKYlike Hamiltonian of the MI spins: with the RKKY energy scale This anisotropic spin coupling is ferromagnetic for all distances.Note that the RKKY coupling in a normal metal also starts out with a ferromagnetic sign at short distances [43].The first two expressions in Eq. (20b) are valid in the wide-band limit.In the presence of a finite but large bandwidth, this RKKY scale can also be rewritten as in the last expression where the RKKY coupling in units of the bandwidth (D) is simply given by the dimensionless j 2 0 [Eq.( 6)] p q X V 7 j Z F 5 e q z V 8 e W f z R 9 c i X / y y 9 z o f f T D r 7 / k D P w p D P 8 I W B F F 8 G H e A j K J w h M n A X 9 X e 8 c H H D 1 X v 9 t P Z r P / Z O r O 6 4 A q Y p M Z M i F / B t K E a B J O 8 9 Z L a 8 I q y K 5 r z i Y W K F t x M m 1 U w L X 5 h m T n O S m 2 X A r x i 7 0 4 0 t D B m W a S 2 s 6 C w M P 9 q H f k / b V J D d j R t h K p q 4 I q t D 8 p q i a H E X c p 4 L j R n I J c W U K a F v S t m C 2 q z B f s X 3 i q E 0 Z D E Z G j f P g o D 3 w 8 t i C N y S I 5 + h 3 A e D M h w E L y x a Z y g d f X Q L n q O 9 h F B M T p G r 9 E Z G i O G v q D v z o a z 6 X x z k b v l b q 9 b X e f n z D P 0 V 7 k 7 P w C X B b e 3 < / l a t e x i t > \|ui = \|""i < l a t e x i t s h a 1 _ b a s e 6 4 = " / e 3 t x 1 J q e D + p 9 k W M g q B 4 N S C + d 6 n J X Y n w q L S m q 4 r i W V g 1 L I S z G C n q d G 5 O D 6 0 / l 1 1 / T Q K 0 O a F d Y / g 3 S u P p y Y i t y 5 S Z 7 6 Z C 5 w 7 J 5 6 M / E 5 r 1 d h d t y f K l N W C E Y u F m W V p l j Q W V V 0 q C x I 1 B N P h L T K / 5 X K s b B C o i 9 0 U U K 7 x W P e 8 r e 3 w y Z j o S d x x I / 4 8 f 8 S z p o N 3 m o 0 P / o 2 3 p E F 1 s l r s k / e E E 5 i 0 i E f y C n p E k l u y T f y g / w M v g b f g 1 / B 7 0 V 0 K b i f e U U e I f j 7 D y K M p c Y = < / l a t e x i t > \|di = \|##i < l a t e x i t s h a 1 _ b a s e 6 4 = " t / N z m w 0 l p Q e S p h u P 0 9 O v c P z 4 D 6 1 R h v u O 0 h E E u x k Z l S g r 0 0 r D e v h g l V p i x B r p H a a I h w w u a j I p z I 6 w t z p 9 R q 8 Y T f E z X h v U d 1 m B R G L K I e t K M 4 t 1 4 T n g n C j u U N 9 g C O 9 3 t 5 M u P u + 7 0 c F i / 9 X f J K g e D U g v n + p y V O J g J i 0 p q u K w l l Y N S y F M x h r 6 n R u T g B r P F h p f 0 k 1 d G N C u s P w b p Q v 1 7 Y i Z y 5 6 Z 5 6 p O 5 w I n 7 1 5 u L / / P 6 F W b t w U y Z s k I w c v l Q V m m K B Z 3 X R U f K g k Q 9 9 U R I q / x f q Z w I K y T 6 U p c l d F o 8 5 i 2 / e y d s M h Z 6 E k d 8 l 7 e f S j h q N n i r 0 f z m 2 9 g n S 6 y T j 2 S b f C a c x K R L D s g h 6 R F J f p J r 8 p v 8 C a 6 C X 8 F N c L u M r g S P M 1 v k G Y L 7 B 2 1 o p 4 M = < / l a t e x i t > E FIG. 2. Eigenstates and eigenlevels of the RKKY Hamiltonian (20).Red lines and arrows show various decay channels of the singlet state (dashed lines) and up/down states (solid lines).These include spin-flips (outer paths) or phase flip (center path), as probed by the transverse ⟨S + L ∥S − R ⟩ω etc. and the longitudinal ⟨S z L ∥S z R ⟩ω dynamical spin correlation function, and which in the helical context reflect backward and forward scattering, respectively.divided by the distance of the impurities in units of the lattice spacing [Eq.( 2)].The wideband limit in the analytical approach therefore implies two assumptions: based on the second order approach used to derive Eq. (20b), this implies (i) j 0 ≡ ρ 0 J ≪ 1, i.e., J ≪ D. Yet via Eq.(2) [see also Eq. (D7)], the wideband limit also implies (ii) x ≫ a. The ground state of this MI spin Hamiltonian is the triplet state with S z =0 (cf.Fig. 2): Hence despite the ferromagnetic coupling in Eq. (20a), the spins are antialigned.Yet by not being in a singlet state, they are still also exposed to Kondo screening.The forward-scattering (∼J z ) parts of the electron-MI coupling do not give a contribution ∼ S z 1 S z 2 to the secondorder effective spin action.This is because the contribution of the S z 1 S z 2 -term to Eqs.(15-16b) contains the product of two Green's functions of the same chirality, G σ (x, iω n )G σ (−x, iω n ), which vanishes in Eq.(16b) due to Eq. (11).Cross-scattering of the type S z S ± is also absent in the effective second-order spin action in Eq. ( 18) because, in equilibrium, the electrons of different chiralities are not correlated.Namely, the averaging of the corresponding combinations of fermion operators contain three operators of the same chirality, e.g. )⟩ which vanishes due to the property ⟨Ψ L (X 1 )Ψ † R (X 2 )⟩ = 0.In addition to the ground state \|t⟩, the effective spin Hamiltonian (20) possesses three other eigenstates with higher energies (cf.Fig. 2): the remaining triplet states denoted as the degenerate doublet ('up' and 'down') states, and the singlet state, The excitation energies of the doublet and the singlet relative to the ground state are E R and 2E R , respectively (cf.Fig. 2). There is an important physical difference between the ground and all excited states of the effective spin Hamiltonian.The first one is stable and corresponds to the ground state of the total many-particle (electrons + spins) system projected onto the spin sector.To be specific, in the wideband limit D → ∞ (i.e., j 0 → 0) and zero temperature, the pair of impurities live in an exact non-correlated product ground state with the helical edge channel, i.e., \|g⟩ ≡ \|t⟩ ⊗ \|0⟩ edge .In contrast, the excited states of the spins remain connected to the many-electron "reservoir" and as such cannot be simply written like product states for true eigenstates of the entire system.Instead, they represent resonant states, in the sense that they give rise to (narrow) resonances in the dynamical spin response functions. By using the Fermi Golden rule, one can estimate decay rates of the excited spin states: the decay rate of the states \|u⟩ and \|d⟩ is ∝ J2 0 E R while the state \|s⟩ has the parametrically smaller decay rate J2 z J4 0 E R (see [44] for more details). B. Taking into account non-perturbative effects of electron interactions or finite Jz Non-perturbative effects of the electron interactions and of a finite J z on the indirect exchange interaction of two MIs attached to the helical edge can be described by using the bosonized theory.The standard free (without impurities) action reads as: The single bosonic field ϕ describes both the spin and chiral degrees of freedom [16,21].K and u are the Luttinger parameter and speed of the helical plasmons, respectively.K incorporates effects of the electron interaction.The boson-spin exchange interaction is described by actions: × S + j e −2iϕ + c.c. . Here, subscripts fs and bs stand for forward-/backward scattering, and ξ is the lattice constant which is usually needed to make the bosonized theory regular.By using the Emery-Kivelson gauge transformation [45], one can completely reduce the effect of J z to changing the dimension of the backscattering, which can be described by the effective dependence of K on J z : Below, we assume that Jz is included in K. The bosonic theory is not quadratic and a functional integral over the bosonic fields D{ϕ}e −(S b +S bs ) (28) cannot be calculated exactly, even formally.However, the effective action of the indirect spin interaction can be obtained for small J by calculating the integral over ϕ as the first cumulant.This is similar to the renormalization group (RG) treatment of the sine-Gordon theory [40]: one Taylor expands the exponential in Eq.( 28) in S bs up to the second order, calculates the integral over ϕ and re-exponentiates the answer: where ⟨A⟩ S b ≡ D{ϕ} Ae −S b .Using the well-known expression for the bosonic correlation function we arrive at the perturbative (in J) expression for the effective spin action This theory is non-local in time and, thus, takes into account retardation effects which are beyond the Hamiltonian formulation of the RKKY theory.The action can be reduced to a local one if ξ ≪ \|x 1 − x 2 \| ≪ L T and 1/2 < K ≤ 1.In this case, Eq. ( 30) reduces to These expressions were analyzed in Ref. [1].They coincide with answers derived in the previous section for the noninteracting case, K = 1. The application of the bosonized theory has several advantages.Firstly, it takes into account non-perturbative effects of the electron interaction and of J z .Besides, it is straightforward to go beyond the quadratic approximation in J and derive renormalization of this Kondo coupling constant [16,21]. IV. NRG ANALYSIS OF THE 2HKM We proceed by presenting the NRG results for the 2HKM in this section, before we explain in detail how we setup the two-impurity helical setup within the NRG framework in the subsequent Sec. A. An exhaustive NRG parameter scan for the 2HKM at J = 0.1 is shown in Fig. 3.All panels have the anisotropy parameter J z /J on their vertical axis whereas the horizontal axis shows energy in various forms [temperature [first column, i.e., Figs.3(:,1)], NRG energy scale [second column Figs.3(:,2)], or frequency (remainder of columns, Figs.3(:,3-5))].Throughout, energy is decreasing towards the right, as motivated by the NRG approach (second column), where large energy scales come first, followed by a zoom into exponentially small energy scales towards the right.Each row labelled by a letter shows data for a fixed impurity distance x as indicated in the left panel.This distance is always chosen integer, i.e., on the grid (A11), and increases exponentially towards lower panels, with the value specified in the left panels. The inverse time to travel in between the impurities naturally gives rise to its own energy scale which may be interpreted as the 'Thouless energy' of a quantum dot confined by the magnetic impurities.This energy scale in itself is independent of any impurity properties other than their distance [see also Eq. ( A9)].Now since the impurity distance explicitly enters the construction of the Wilson ladder (A34), there is always a clear qualitative change in the NRG finite size spectra, aka.energy flow diagram at the energy scale E x , as visualized in a condensed graphic way in Fig. 3(:,2), i.e., the second column.There one observes a distinct change (whitish to gray transition) right at E x (vertical dotted line).This 'curtain' is opened, i.e., moves towards the right as the impurity distance is increased from the top to the bottom panels.As such this 'unveils' the underlying Kondo physics.At largest distance shown in the bottom row of panels, Fig. 3(h,:), E R has already dropped below the smallest Hence in this case the impurities are fully Kondo screened individually, and RKKY physics plays no role any longer, and hence is absent for all J z ≥ 0. At larger energies above the coherence scale, E ≳ E x , i.e., to the left of the vertical dotted black line, the system is described by effectively independent impurities.Since information cannot travel faster than v between the impurities, the impurities do not yet 'see' each other at energy scales E > E x .In this sense, the finite size spectra in the NRG look identical in Fig. 3(:,2) for large energies E ≳ E x , i.e., to the left of vertical dotted line across all panels in Figs.3(:,2). For low energies, Fig. 3 shows that the smallest energy scale from the point of the impurity is max(E R , T K ).Therefore E R serves as a low-energy cutoff.For example, the energy flow diagram is converged below E R [uniform gray area to the right of E R (blue vertical marker) in Figs.3(:,2)].On explicit physical grounds this is replicated in terms of static inter-impurity correlations for T < E x in Figs.3(:,1), or by having no particular structure in the spectral density for \|ω\| < E R in the right panels.For this reason the brown line which depicts the Kondo scale, is only shown above E R and hence terminated at the blue vertical marker line.For J z above this crossing point, Kondo screening sets in and eventually fully dominates.This is seen in the brightening of the dark red inter-impurity spin correlations in Fig. 3(:,1) towards white (no correlations) when increasing J z (vertical direction).(for precise prefactors, see [46]).This encodes cumulative effective thermal weights in three arbitrary but fixed low-energy symmetry sectors into red-green-blue (RGB) colors for all odd Wilson shells n (to avoid even/odd effects) based on an effective inverse temperature βn ≡ 4ωn.The symmetry sectors chosen for this were q ≡ (Q; S tot z ) ∈ (0; 0, 1, −1) with Q the total charge relative to half-filling.The remainder of the columns shows dynamical spin-spin correlation functions ⟨ Ŝη ∥ Ŝη ′ † ⟩ω as indicated at the top of each column at zero temperature (T = 1.8 10 −11 ).The solid-dotted lines shows the respective derived inverse static susceptibility T ηη ′ S = 1/4χ ηη ′ S [cf.Eq. ( 34)].Blue (red) indicates positive (negative) value, respectively.The number at the top right of each panel indicates the maximum absolute value the spectral data A of the broadened NRG spectral data, and hence gives an impression of numerical range.Blue (red) shading in all panels except for the second column indicates positive (negative) values, respectively.The dashed lines in (a,c;3) and (a,c;5) represent exponential fits as indicated with (a5) of the maximum of the spectral data for J/Jz ≤ 0.5.NRG parameters (e.g., see [46] for detailed definitions): Λ = 4, truncation energy Etrunc = 8; z-averaged over nz = 4, with log-Gauss broadening σ = 0.3 of the spectral data after z-averaging.2)], the dimensionless Kondo coupling in Fig. 3 is small, having j 0 ≡ ρ 0 J = 0.05 ≪ 1, and thus Ex ER = 1 πj 2 0 = 127.3.Therefore the bare RKKY regime spans about two orders of magnitude in energy scale.It competes with Kondo physics for the case T K (J, J z ) < E x , i.e., when the Kondo scale is small enough that the screening clouds of the two impurities overlap.We refer to this intersect as the intermediate energy regime. The bare RKKY regime concerns the energy window This intermediate regime becomes visible as the lighterblue shaded area in Fig. 3(a-f,2) in between the two vertical makers below the brown solid line which represents the single-impurity Kondo scale T K .The latter scale T K is cut off at E R , also visually so by terminating its line towards the right at the blue vertical marker.Hence the intermediate regime is present (i) if there is a brown line segment in between the two vertical markers (black dashed and blue), in which case (ii) the intermediate regime occurs below it.For example, with no brown line segment in the bare RKKY regime in the lowest two rows in Fig. 3, T K > E x dominates the low-energy regime throughout, and the system consists of two individually Kondo screened impurities.However, once T K < E x , the buildup of the Kondo screening cloud is affected by the presence of the other impurity.This leads to characteristic changes in the NRG energy flow diagrams, as seen in Fig. 3(:,2). A. Spin-spin correlation functions and low-energy scales Data for dynamical spin-spin correlation function A(ω) [as defined in Eq. (A53) in App.A 4] is shown as color-plots in the right panels of Fig. 3.The frequency of the maximum for fixed parameters sets the relevant low-energy scale for the spins.It is well traced by the inverse static susceptibility (solid-dotted lines) as derived from the dynamical susceptibility with S ∈ {S z , S ± }, and the normalization convention for In case that η = η ′ , i.e., inter-impurity correlations, only a single label may be shown, e.g., (the η label may be skipped altogether then, since the impurities are considered identical, hence by symmetry, e.g., . Strictly speaking, the interpretation as an energy scale is only justified for 'diagonal' correlations, (here intra-impurity η = η ′ ), as this guarantees positive spectral data and hence a respective positive energy scale.But it is useful to also include η ̸ = η ′ here for the sake of the argument and presentation.We will also refer to χ ηη ′ S± which involves a spin-flip, as the transverse susceptibility, and χ ηη ′ Sz which involves a phase flip, as the longitudinal susceptibility.The choice of the prefactor (1/4) is motivated by the standard definition of the Kondo temperature in the plain single-impurity Kondo model, T NRG K ≡ 1 4χ0 [33,47].For isolated RKKY impurities with the energy spectrum as in Fig. 2, the spectral functions have δ-peaks at energies ω = ±E R for the transverse, and ±2E R for the longitudinal correlation function, thus resulting in T ηη ′ ;0 S± ≡ E R /2 and T ηη ′ ;0 Sz ≡ sgn(ηη ′ )E R , respectively [note the normalization convention of the spin operators as indicated with Eq. ( 35)].Up to a sign, these are independent of the choice of η and η ′ , i.e., they become the same for inter-and intra-impurity susceptibilities.The latter is a direct consequence of the RKKY low-energy regime where the pair of impurities, even though spatially separated, act like a nearly-decouled microscopic unit governed by the RKKY Hamiltonian.In the following we will thus scale energies in the numerical data by the smaller energy scale (where subscript 'S' denotes 'spin'), We prefer T S over E R , since E R only represents a lowest order estimate, whereas T S includes the full many-body aspects of the problem and is thus also self-contained and thus consistent within the NRG. The energy scales T ηη ′ S in Eq. ( 34) capture the low-energy scale of the impurity spins, as seen, e.g., in the lowest panels Figs.3(f-h,3), [with the impurity operators S as well as their location η and η ′ specified at the top of each of the right columns Figs.3(:,3-5)].There the inverse susceptibility T S± from the intra-impurity spin-spin correlation (soliddotted line) [Figs.3(:,3)] follows closely the analytical Kondo scale T K (J, J z ) (brown line) up to a constant prefactor of order one.This generally holds for ω > E x , i.e., to the left of the black dashed line in Figs.3(:,3). The two right-most columns of Fig. 3, in contrast, show inter-impurity correlations.These can only be due to RKKY interactions, and hence diminish with increasing impurity distance.Once this distance exceeds the size scale of the Kondo screening cloud, i.e., T K > E x , the inter-impurity susceptibility becomes much smaller than the onsite susceptibility, such that its inverse is orders of magnitude larger in energy [e.g.compare solid-dotted line in Figs.3(gh;4,5) to T K (brown line)].Its physical interpretation is that inter-impurity correlations start to play a role relative to Kondo correlations only once the latter are sufficiently suppressed, e.g., by a large temperature scale T ≳ T LR S± ≫ T K .In the intermediate regime, where the low-energy physics is cut off by RKKY, the inter-and intra-spin correlations start to look identical when applying the same operators [e.g., compare Fig. 3(a,3) to Fig. 3(a,4)].This holds quantitatively as also seen by the overall scale (see maximum spectral weight A indicated with the panels).This can be understood based on the RKKY impurity state \|t⟩ in Eq. ( 21) that (nearly) decouples as a product state from the bath channel [cf.App.A 7], thus having which holds for both, η = η ′ (intra-impurity) as well as η ̸ = η ′ (inter-impurity), while bearing in mind that the equal-time correlator above is identical to the integrated spectral data over frequency [cf.sum rules]. The longitudinal inter-impurity correlations are shown in the last column of Fig. 3. Deep in the RKKY, the dominant spectral weight of this non-diagonal dynamical correlation functions is expected to be negative, and half the absolute value, in agreement with the red shading (which indicates negative) and overall scale of the spectral data, e.g., in Fig. 3(a5) [see a detailed analysis of the effects of finite bandwidth on the precise value of the l.h.s. in Eq. (38) in App.A 7, and in particular Fig. 11 therein].Via spectral sum rule, the frequency integrated data yields the data in Fig. 3(:,1) at given temperature (in the present case, at the lowest temperature shown).Consistently, this appears in a deep red in the RKKY state, indicating that the spin-orientations of the two impurities are antiferromagnetically (AF) correlated.However, when crossing over into the Kondo regime, the weakened inter-impurity longitudinal correlations can even change sign and turn ferromagnetic (FM) [e.g., blue soliddotted lines in Figs.3(:,5) in the inverse susceptibility; this necessarily has to originate from corresponding positive spectral data, but its weight is too small, though, so it is not visible in shading in the spectral data as shown].The weak ferromagnetic correlation can also be partly seen in the static equaltime spin correlation [e.g., the very faint blue hue in the intermediate regime just left of the red area in Fig. 3(b,1) or (e,1), and hence at significantly elevated temperatures].The sign change towards weak ferromagnetic correlation observed in the longitudinal data is not systematic, though.For example, not all inverse susceptibilities in Figs.3(:,5) show a sign change.Furthermore, deep in the Kondo regime, the system can feature weak ferromagnetic (FM) inter-impurity spin correlations and together with a sign change depending on the frequency [e.g., faint red and blue shadings in Figs.3(g-h,5) just above J z = 0 vs. ω, which is partly also visibly replicated vs. temperature e.g., in Figs.3(g,1)].For energies below E x this should be well resolved by NRG, and not related to coherence affects averaged out by NRG above E x due to coarse graining [cf.discussion following Eq.(A9)].The appearance of weak AF as well as FM correlations across the impurities may be related to the fact, that for finite J z and finite bandwidth, subleading terms can generate an effective small longitudinal inter-impurity interaction J z ŜL z ŜR z which due to its oscillatory behavior vs. distance may be ferromagnetic as well as antiferromagnetic [cf.App.A 7, and also the discussion around Eq. (D11)].Consequently then, z-averaging of the spectral data may lead to apparent non-systematic behavior in the longitudinal correlations as a numerical artifact.In the present case, however, we do not dwell on this any longer as this concerns subleading effects. B. RKKY scale in spectral data The inverse transverse susceptibility deep in the RKKY regime resembles a straight line on a semilog plot [Fig.3(a, [3][4]].This also reflects the behavior of the maximum in the actual spectral data.Generally, from its very definition via the Kramers-Kronig integral relations, the inverse susceptibility is also sensitive to the precise line shape of the spectral data.Yet by having the maxima and inverse susceptibility run in parallel for small J/J z , this suggests similar line shapes.Tracking and fitting the peak maxima in the spectral data by the exponential fit specified with Fig. 3(a5) for J/J z ≤ 0.5, we obtain ω * ≃ 2.78•10 −4 e 0.82Jz (blue dashed line).It is lower-bound at J z = 0 by the analytically obtained RKKY scale E R [Eq.(20); blue vertical marker in Fig. 3], having E R = (ρ 0 J) 2 /x = 2.50•10 −4 for x = 10 [Fig.3(a,:)].The difference of about 10% is due to finite bandwidth [cf.App.A 7]. As seen from the fits in Fig. 3(a;3,5) the low-energy scale in the intermediate regime diminishes exponentially with decreasing J z , as in T S ∼ ω * ≃ ae bJz with b > 0. This is qualitatively similar to the one-impurity Kondo temperature in the anisotropic 1-impurity case (cf.App.B) yet with different renormalized coefficients.For one, this shows that the RG / poor-man's scaling for a single impurity needs to be stopped at the RKKY scale.Moreover, the relative slope in the exponents of the low-energy scale as seen in the semilog plots in Figs.3(:,4) changes in the intermediate regime with increasing impurity distance.The slope 1/b of the peak vs. ω * is larger than for T K (brown line) in Fig. 3(a;3,4), about comparable in Fig. 3(e,4), and smaller in Fig. 3(f-h;4) where RKKY is absent.This clearly underlines the continuous crossover from RKKY to Kondo. A similar exponential fit on the maximum of the spectral data was also obtained for the longitudinal spin correlations in Fig. 3(a5).The peak in the spectral data occurs at a larger a ′ = 5.86•10 −4 at J z = 0 when compared to the corresponding fit in Fig. 3(a3).The slopes b are comparable, though (b ′ = 0.80 vs. b = 0.82).Therefore the maximum in the longitudical spectral data ⟨ ŜL z ∥ ŜR z ⟩ ω is systematically shifted at small J z by a factor of 5.86/2.78= 2.11 towards larger energies as compared to the maximum in the spectral data that requires a spin-flip, ⟨ ŜL + ∥ ŜR − ⟩ ω .Thus in the RKKY regime at x = 10 [Figs.3(a,:)], and consistent with the explicit analytical expression for E R in Eq. ( 20) This is in agreement with the effective twoimpurity level-spectrum in Fig. 2, where with reference to the Lehmann representation for spectral data in Eq. (A53), one needs to pay an energy E R for a spin-flip, whereas one needs to pay an energy 2E R for a sign-flip (triplet \|t⟩ to singlet \|s⟩ transition).The deviation of about 5% from the expected factor of 2 in the fitted values is within the spectral resolution of NRG, and thus likely due to z-averaging and broadening of the spectral data.The parameter scans in Fig. 3 give important hints: they show that E R (blue vertical marker in Fig. 3) as obtained from a second order perturbative approach [cf.Eq. (20b)], does not always describe the RKKY low-energy scale correctly, as it can get renormalized by the presence of single-impurity Kondo correlations.While for x = 10 this well coincides with the J z = 0 low-energy scale [e.g., compare to solid lines with symbols in Fig. 3(a,4)], when increasing the distance, the peak in the NRG data shifts towards the left of the vertical blue marker, i.e., towards values that are larger than E R .That is, the fit value for a shown with Fig. 3(a;3,5) effectively increases relative to E R when exponentially increasing the distance x (no additional fits shown, though, as this is a qualitative argument).Thus even if at J z = 0 the oneimpurity Kondo scale may still be many orders of magnitude smaller than E R , the value of E R already gets weakly affected (likely acquires logarithmic corrections) due to the underlying Kondo correlations, even if T K ≪ E R .If at the same time it also holds E R ≪ D = 1, then integrating out the helical edge mode starting from the initial band edge D towards zero energy, e.g., in a poor-man's scaling sense, this can be expected to introduce an RG flow also for E R .This suggests that for a consistent interpretation of the NRG data with analytics, the NRG data should be scaled by T (η) Sz [cf.Eq. ( 34)], rather than the lowest order analytical estimate E R in Eq. (20b). Moreover, the peaks seen in the spectral data are rather narrow, i.e., within the resolution limit of the presented NRG data.Based on the chosen NRG parameters Λ = 4 with n z = 4 z-shifts (e.g., see Sec.A 3, or also [46] for more detailed definitions), we expect a best possible relative spectral resolution δω/ω > Λ 1/2nz − 1 = 0.19, hence the value used for the broadening of σ = 0.3 on the discrete NRG raw data.Based on perturbative approach, however, one may suspect significantly narrower features as shown in Fig. 3.With this in mind, the spectral features seen in data in Fig. 3 are likely overbroadened. C. More detailed spectral analysis (line shapes) The RKKY 'resonances' in the spectral data are expected to be potentially very narrow.As it turns out, though, while sharp peaks (near delta spikes) can be found in the bare discrete spectral NRG data, the precise location in frequency of these peaks is sensitive on the z-shift in the logarithmic discretization of the NRG setup (e.g., see Sec.A 3 or also [46] for detailed definitions).Specifically, as the z-shifts can shift energies by a factor of √ Λ in the discrete setup, in the present case, also the respective 'response' of the system in terms of the precise location of a narrow spectral resonance for the 2HKM model can also vary within a factor of √ Λ. Therefore blind z-averaging of the NRG data leads to artificial broadening and somewhat irregular z-averaged data, as demonstrated in Fig. 4 or Fig. 6 for two different impurity distances, x = 10 and x = 1 000, respectively.Instead, when scaling the spectral data for each individual z-shift by its respective T Sα (z) and averaging that resulting data, peak shapes are significantly improved, and in particular also narrower (e.g., compare Fig. 4 → 5, or Fig. 6 → 7). For the analysis here, the rather large Λ = 4 is useful to emphasize how discretization manifests itself in the spectral data which is more subtle here, as it also leads to shifts.A major motivation for the larger Λ = 4 is, of course, that this FIG. 5. Exactly the same bare data as in Fig. 4, except that for each z-shift the frequency scale is first scaled by T ηη ′ Sα /T ηη ′ Sα (z) (with T ηη ′ Sα ≡ ⟨T ηη ′ Sα (z)⟩z the z-averaged value, and α ∈ {z, ±} chosen, respectively, for each curve), and then combined as in Fig. 4. The zshift specific TS z (z) is also determined within the NRG in the same calculation as part of the post analysis [cf.Eq. ( 34)].Overall, this procedure leads to a significantly improved quality of peak shape that is significantly narrower, as compared to the spurious spread of spectral peaks in the right panels in Fig. 4. also results in faster NRG calculation or, conversely, in better converged NRG spectral data.By careful z-averaging, we can obtain good spectral resolution for Λ = 4, nevertheless.However, given that RKKY peaks can be expected to become much narrower in terms of width vs. location, this ultimately will be also challenging for smaller Λ in any case.Overall, the FIG. 6. Same analysis as in Fig. 6 except for an increased impurity distance x = 1 000, leading to the reduced value for TS z = 3.0•10 −6 = 1.19 ER.The broadening was also increase to σ = 0.15.FIG. 7. Exactly the same bare data as in Fig. 6, except that for each z-shift the frequency scale is first scaled by T ηη ′ Sα /T ηη ′ Sα (z) (similar to Fig. 5). present analysis in terms of Λ = 4 already clearly supports narrow features.More importantly, the location of the peaks, and hence the corresponding energy scale, can be considered reliable and significantly more accurate than the width, given also that these are directly related to inverse static susceptibilities. The data in the right panels of Fig. 5 gives a good impression for the spectral data in the RKKY regime.As compared to Eq. ( 36) with T S = 0.50 E R , the actual data in Figs. 4 & 5 gives T S = 0.556 E R which is reasonably close.Furthermore, the peak location in the spectral data is expected at ω * = E R ≃ 2T S for the transverse spectral data, and at ω * = 2E R ≃ 4T S in the longitudinal data, again consistent with the data in Fig. 5.The peak width in the longitudinal data related to a phase flip when transitioning to the singlet state [cf.Fig. 2] is found to be comparable as in the transverse data [44].However, this width is limited by the broadening σ as seen with the individual data in the left panels.While some structure can be observed with Fig. 5(a) resolved by the spread with z-shifts, the data in Fig. 5(b) is more smooth this way, just showing the broadening σ = 0.10 used.This suggests that the data in (b) is still likely overbroadened by given σ = 0.10.Finally, as already argued with Fig. 3, the inter-and intra-impurity correlations are identical to each other deep in the RKKY regime.Here this is seen by having the dashed lines in the right panels of Fig. 5 on top of the solid lines [note the sign change, though, as indicated with legend in Fig. 5(d)], which again is rooted in the triplet ground state in Eq. (21). Repeating the same analysis for the increased impurity distance x = 1 → 1 000, the data in Fig. 7 starts to show a qualitative change in the spectral line shape.While having increased the broadening to σ = 0.15, as reflected by the individual peaks in the left panels, the overall lineshape in Fig. 7(d) is still largely comparable.The interand intra-impurity correlations do lie nearly exactly on top of each other, thus also suggesting a clear RKKY low-energy state still.In particular, peak position are still at the expected ω * = 2T Sz or 4T Sz for the transverse or longitudinal data.The numerical value for T Sz , however, further deviates from the plain second-order perturbative RKKY scale, having T Sz = 3.0•10 −6 = 1.19 E R which by now is clearly unequal from the naive expected value of 0.5 E R (see also Fig. 8 for a more detailed analysis in this regard).Yet when scaling the data consistently fully within the NRG framework, the schematic picture in Fig. 2 still works well when substituting To conclude this section, we reemphasize that the NRG approach correctly describes the position of the peaks in the response functions as well as the integrated spectral weight, but yields resolution-limited information about the peak width and thus its overall shape.The latter unavoidably results from the coarse-graining in energy space intrinsic to the NRG approach. D. RKKY vs. Kondo Energy scales An analysis of the low-energy scales vs. impurity distance is shown in Fig. 8.The longitudinal inter-impurity susceptibilities are negative due to the antiferromagnetic RKKY correlations; note the sign with T LR Sz in the legend.In the crossover region when leaving the RKKY regime, this can become positive, though [cf.legend with Fig. 8(c)].As seen in all panels, the energy scale T S 'drifts' away from the bare RKKY scale E R (black dashed line) already many orders above the Kondo scale T K (indicated or specified with the upper panels).If the distance exceeds the temperature length scale v/T (left panels) or the inverse Kondo scale (right panels), the impurities effectively become independent of each other, such that RKKY is cutoff by max(T, T K ).In Fig. 8, the end of the RKKY regime is seen where inter-impurity correlations start to deviate significantly from their intra-impurity counter part (e.g., compare blue vs. yellow in the lower panels).By approaching the wide-band limit (or equivalently, by reducing J, as in right vs. left panels in Fig. 8), one can observe in the lower panels that T Sz ≃ 2(T S± ≡ T S ) is obeyed over a wide distance (energy) window.Also deviations from the relative factor of 2 diminish towards the wide-band limit.The shad- ing in the lower panels of Fig. 8 indicates variations (i.e., the standard deviation) due to by z-shifts in the logarithmic discretization.These also become smaller towards the wide-band limit [e.g., compare Fig. 8(d → b)]. V. SUMMARY AND OUTLOOK We presented a broad parameter scan of the two-impurity helical Kondo model (2HKM) vs. impurity distance, coupling anisotropy, temperature, and finite bandwidth within the NRG framework.We emphasize that this setup substantially differs from a chiral edge mode, where both spins propagate in the same direction [for a recent NRG study on the latter, see e.g., Lotem et al. [41]; the chiral model is different from the helical edge mode discussed here, both in the setup as well as in the physics].With the NRG being non-perturbative in character, our presented results are reliable quantitatively in the full parameter regime, with the only real constraint being energy resolution in spectral data.We have established a plain crossover from RKKY to Kondo with increasing impurity distance or, conversely, increasing Kondo coupling or its anisotropy.The Kondo screening of the impurities individually by the helical edge mode is tuned continuously into an effective mutual 'RKKY' screening of the impurities themselves.In this sense, the Kondo renormalization flow is cutoff by the RKKY energy E R once it exceeds the Kondo scale.The RKKY occurs in the energy window determined by the inverse time scale required for the impurity to travel in between the impurities (Thouless energy E x ), and the actual RRKY energy E R ≪ E x .If the Kondo scale falls within this window one comes across a continuous crossover. The low-energy effective RKKY Hamiltonian gives rise to narrow resonances with the helical edge.While the presented NRG analysis, the location and overall spectral weight is reliable, the spectral width, however, is likely (much) below the energy resolution of our NRG data deep within the RKKY far from any Kondo screening.Hence the precise linewidth e.g., in the dynamical spin-spin correlation data for the fully interacting model is left for future analytical and numerical studies. Hybridization function The helical 1D edge mode above is assumed to constitute a fermionic macroscopic bath that interacts equally with two impurities dησ symmetrically located at positions x η = η 2 x with η = ±1 [Eq.( 5)].The NRG approach first focuses on Anderson type impurities with explicit hybridization of the impurities with the bath.The switch to Kondo-type impurities can be taken as a subsequent step, e.g., via Schrieffer-Wolff transformation.Thus the starting point is the hybridization Hamiltonian, This defines the bath states f0ησ 2 ĉkσ that the impurities couple to, which permits the interpretation that these are the bath states at the respective location of the impurities.The states f0ησ are normalized, but not necessarily orthogonal yet, hence the tilde (cf.Sec.A 2). In Eq. (A2), an electron with spin σ can hop on and off the corresponding helical branch, with the spin σ preserved in the process.The hopping amplitude V k is assumed independent and thus symmetric in the spin and impurity indices.For simplicity, the bath is also assumed featureless, characterized by two parameters only: a hybridization strength Γ and a finite half-bandwidth D. In the continuum limit, the hybridization function for each impurity individually is given by Γ(ε) = πρ ε \|V ε \| 2 ≡ Γ ϑ(D − \|ε\|) (aka.the default NRG box distribution), with ρ ε = ρ 0 ϑ(D − \|ε\|) the one-particle density of states, assuming constant ρ 0 = 1 2D without restricting case.The hybridization is cut off sharply in energy, as depicted in Fig. 1.The integrated (norm-squared) hybridization strength which yields the splitoff normalization factor in the second line in Eq. (A2). From the perspective of the impurities, the full hybridization function becomes a 2 × 2 matrix with η ∈ {R, L} indexing the impurities [Eq.( 5)].This includes a non-zero off-diagonal hybridization for η ̸ = η ′ .In the wide-band limit D → ∞, the hybridization becomes where τ ≡ σx v is the time required for a particle to travel from one impurity to the other, since for example from Eq. (A3) for a particle to travel from L → R (thus with x ̸ = 0, also τ ̸ = 0), This gives a non-zero contribution only for τ > 0. Assuming v, x > 0, for example, this requires σ > 0. Indeed, in given helical setting, a spin up electron travels to the right, that is the particle needs to be created at the left impurity (η ′ = −1) which then can propagate to the right impurity (η = +1).For spin down, the situation is vice versa. We emphasize that the spectral representation Γ(ω) of the hybridization function cannot be simply written as −Im∆(ω) in the present case because the matrix elements in Eq. (A3) are complex.Instead, the imaginary part needs to be taken from the propagator only, i.e., with − Im where the spectral function Γ ηη ′ (ω) ≡ [Γ(ω)] ηη ′ needs to be differentiated from the constant Γ.Correspondingly, in matrix notation, which is non-zero now for both off-diagonal entries η ̸ = η ′ for either spin σ.Only by taking the spectral data as above, this can be simply completed to the full hybridization function using Kramers-Kronig transform on the complex spectral data, i.e., by folding the above possibly complex spectral function with 1/(ω + − ε).Within the wide-band limit and in the absence of interactions, the Green's function for the impurities with onsite energies ε dη becomes, e.g., in the spin-up channel (σ = +1, and hence τ > 0), At particle hole symmetry with ε dη = 0 on obtains for ω = 0, where the diagonal entries reflect half-filling based on the Friedel sum rule.Due to the helicity, the diagonal terms in Eq. (A7) are just the Green's function of decoupled impurities, G ηη (ω) = 1 ω−ε dη +iΓ .The non-zero off-diagonal term maintains the same matrix position as in ∆(ω) in Eq. (A4a), consistent with the directedness of propagation.Similar to Eq. (A4a), the off-diagonal term also shows oscillatory behavior in energy with period This period is fixed by the energy scale E x = v/x [Eq.( 33)] set by the inverse time τ required for an electron to travel from one impurity to the other.In particular, the period δω is independent of the energy scale ω in G(ω).Given that the NRG discretizes logarithmically in energy space, assuming E x ≪ D = 1, there is no way these oscillations can be resolved to orders of magnitude higher in energy all the way up to D. On the other hand, on physical grounds at high energies, e.g., at temperatures T > E x , the impurities effectively no longer see each other.In this sense it appears reasonable to expect for most physical quantities such as spin-spin interactions that do not explicitly resolve one-particle phases of propagation.In the presence of relaxation processes due to interactions, these rapid oscillations likely average out at large frequencies even for small temperature, and thus become less important in detail.Yet this needs to be verified in practice on a case-by-case basis by tracking the stability of the data with respect to the level of course graining. Normalization of bath states and effects of finite bandwidth For either spin, the hybridization term Eq. (A2) in the Hamiltonian defines two bath states f0ησ with η ∈ {R, L}.These are normalized, but not strictly orthogonal, hence the tildes with the f operators in Eq. (A2) as a reminder.These bath states at the location of the two impurities have finite overlap that can be determined from the fermionic anticommutator relation, with \|0⟩ the vacuum state, and a ≡ πv D as in Eq. ( 2), and therefore πx a = Dx v = \|τ D\|.Rewritten in matrix notation it can be diagonalized, with eigenvalue matrix with eigenvalues sorted as s 0η = 1+ηr 0 , with η ∈ {+1, −1}, i.e., with the larger eigenvalue s 0+ ≥ 1 coming first.The off-diagonal term in Eq. (A10b), r 0 ≡ a δ a (x), represents a δ-function of width a.In this sense, the UV cut-off D again directly gives rise to the lattice constant already introduced with Eq. ( 2).It has unit of length and resembles the resolution in real space based on the given one-particle density of states encoded into the box distribution.By having a finite bandwidth, this translates into a cutoff in spatial resolution, which in the present case naturally gives rise to a well-defined lattice spacing: the overlap in Eq. (A10) between the two bath states becomes exactly zero for when n ̸ = 0 which naturally suggests a discrete grid with lattice spacing a.As pointed out with Eq. ( 2), choosing D = 1 and a = 1 then fixes the velocity to v = 1/π.For n = 0, i.e., x → 0, the off-diagonal overlap in Eq. (A10a) becomes 1, and thus identical to the diagonal case η = η ′ .In this case the two locations η and η ′ approach the same 'site' and thus become identical. As an aside, we note that the argument here namely that a finite bandwidth naturally gives rise to a discrete lattice spacing, can also be straightforwardly carried over to a plain spinless tight binding chain, with the minor difference that there due to the structured density of states, the Fermi velocity v differs from the value above by a factor of order 1.Furthermore, the Fourier transform of any quadratic Hamiltonian in momentum space yields its full (long-range) hopping structure in real space.Though a purely 1D lattice model is unable to describe the topologically nontrivial phase of the 2D topological insulator, the one-particle dispersion may nevertheless also be Fourier transformed for the isolated single helical edge as in Eq. (1).By starting from momentum space, however, in the present case this would mandate periodic (or infinite) BCs with spin-dependent complex hopping coefficients t nσ ≡ σt n , where with ε kσ ≡ σε k , for a chain of N sites, with distance x = x n = na > 0, having t 0 = 0.These hopping amplitudes are long range, decaying like 1/x, and by construction break inversion symmetry, t n = t * −n ̸ = t −n .The long-range hoppings cannot be eliminated by permitting deviations from the linear dispersion close to the band edge.For example, the altered dispersion εk = 2v a sin ka 2 which still obeys εk ≃ ε k = vk for small k, also results in long-range hopping.This is due to the fact that the one-particle dispersion is discontinuous across the boundary of the Brillouin zone.Given this complications for numerical lattice simulations, the starting point in energymomentum space in Eq. ( 1) is more natural and convenient.Nevertheless, from the above it is clear that the notion of lattice spacing and Brillouin zone are perfectly valid also for a single helical edge mode even if the one-particle dispersion is discontinuous across the boundary of the Brillouin zone. The analytic structure of the overlap in Eq. (A10) is closely related to static fermionic correlations versus distance.For example, consider a filled helical Fermi sea for energies ε ∈ [−D, 0].Then at zero temperature with τ = σx v [Eq.(A4a)], the integral only includes the filled Fermi sea [whereas Eq. (A10a) integrated over the entire 'Brillouine zone'].By comparison with the corresponding Fermionic correlations for plain 1D tight-binding chain, this suggests the Fermi wave vector k f as half the extent of the filled Fermi sea in momentum space, i.e., given half-filling, The leading phase factor in Eq. (A13) has subtle consequences when computing charge correlations, and results in features that qualitatively differ from a plain tight binding chain.Assuming a continuous, i.e., non-discretized 1D edge mode in terms of lattice sites spaced by a, then based on Eq. (A10), the overlap diminishes to zero for x ≫ a.In particular, this includes the wide band limit where a → 0. However, for the sake of orthogonality of the fermionic states, the vanishing of the overlap at finite bandwidth can be simply guaranteed by adhering to the discrete grid in Eq. (A11) with lattice spacing a in complete analogy to a tight-binding chain.Hence, in order to avoid complications based on non-orthogonal f0ησ states, henceforth distances will be chosen on the grid (A11), i.e., with a := 1 having x ∈ Z.With this f0ησ → f0ησ become well-defined orthonormalized local bath modes at the location of the impurities, denoted by using hats now instead of tildes. While finite bandwidth is physically meaningful when having particular 2D lattice models in mind, for a helical edge mode this cutoff is peculiar in that the helical branches merge with a continuum of bulk states.Therefore a sharp ultraviolet cutoff for an isolated 1D helical edge mode can have potentially artificial consequences.Lack of orthogonality of local bath modes discussed with Eq. (A10b) above is one example.The latter complication can be simply eliminated, though, by adhering to the effective discrete lattice in Eq. (A11).On a related footing, the hybridization function in Eq. (A4a), which is closely related to the dynamical one-particle propagation in between the impurities, reflects the directedness of motion via the step functions θ(τ ).This step function, however, is strict for infinite bandwidth only.For finite bandwidth it also contains, in particular, a non-zero oscillatory tail for τ < 0. That is, for \|ω\| ≪ D and τ < 0, rather than strictly being zero, the amplitude for this enhanced backscattering probability decays like 1 D e i(ω±D)τ ∼ 1 D e iωτ [see also Eq. (A6)] where the oscillatory behavior with phase Dτ = πx a is similar to Eq. (A10b), thus having e iτ D = ±1 on the grid Eq. (A11).However, this backscattering probability decays with increasing bandwidth D, which eventually enforces strict directionality.Yet for the above reasons finite bandwidth can generate a weak subleading contribution J z S z L S z R to the RKKY Hamiltonian (20) in the helical system [cf.App.D]. Coarse graining For the sake of a numerical treatment, the continuum of the bath needs to be discretized.Here we use the NRG which, by construction, always discretizes in energy-momentum space.This allows us to target a single edge mode with plain linear dispersion.To be specific, the NRG coarse-grains on a logarithmic grid D/Λ −(n+z) in energy space with Λ > 1 (typically Λ ≳ 2) a dimensionless discretization parameter, n ∈ N, and z ∈ [0, 1[ a plain 'z-shift' of the logarithmic discretization [49,50]. Consider therefore some arbitrary but fixed energy interval I l ≡ [ϵ l , ϵ l+1 ] of width ∆ϵ l ≡ ϵ l+1 −ϵ l > 0 and average energy εl ≡ 1 2 (ϵ l +ϵ l+1 ) within the continuum of the bandwidth.Here l > 0 will refer to energy intervals at positive energies, with energy increasing with increasing l (this is contrary to the NRG, hence l ∼ N − n with N the number of levels with ϵ l > 0).Since the helical mode in Fig. 1 is symmetric around ϵ = 0, the coarse graining for positive and negative energies is also chosen symmetrically around ϵ = 0, having ϵ −l = −ϵ l such that l < 0 corresponds to negative energies.The index l = 0 is generally considered excluded here, as it is typically used to refer to the entire bandwidth, e.g., as with r 0 in Eq. (A10b).Having ϵ −l = −ϵ l , the index l thus resembles momentum, in that the simultaneous inversion of momentum and energy for a given spin flavor leaves the Hamiltonian of the edge mode invariant. The energy εl is differentiated here from ε l (note the different font) with the latter eventually used for the effective level position for the full interval l, typically having ε l ≃ εl similar but not exactly the same [50].When coarse graining the bath, the integral for the hybridization is split up into intervals, with {ĉ εσ , ĉ † εσ }=δ σσ ′ δ(ε−ε ′ ), such that [ĉ εσ ] = energy −1/2 .The coarse-grained discrete and thus dimensionless bath modes clησ are normalized and orthogonal with respect to spin σ, but not yet with respect to the impurity location η, as emphasized by using tildes.The states clησ for each individual interval need to be orthonormalized in any case even if the distance x is chosen on the grid in Eq. (A11).This orthonormalization has to occur prior to the subsequent mapping of the so-called star geometry in Eq. (A15) between the impurity and the bath states to an effective one-dimensional (1D) chain geometry, as the latter requires properly orthonormalized Fermionic levels. Orthonormalization of the pair of bath states within each interval l for given spin can be achieved starting from their overlap which is again simply related to the fermionic anticommutator similar to Eq. (A10), Equation ( A32) is now the starting point for the Lanczos block-tridiagonalization of the bath, given the initial pair of orthonormal states f0 for the zeroth site with real coefficients in the symmetric / antisymmetric basis ĉl of the bath.The bath itself is represented in diagonal form just by the energies of ĉl , and hence is clearly also real.The resulting Wilson chain then consists of two intercoupled chains, with α n , β n ∈ R 2×2 and β 0 reserved for the coupling to f−1 ≡ U 0 d, i.e., the impurities.By choosing the symmetric / antisymmetric basis above, incidentally, it turns out on numerical grounds, that within numerical double-precision accuracy, there are no Creutz-couplings [51], i.e., in between fnη and fn+1,−η .That is, β n turns out diagonal in η, and the Wilson chain becomes a pure ladder (to be referred to as Wilson ladder [52]).Similarly, with the bath setup being symmetric in energy and thus at half-filling, there are also no onsite energies along the Wilson sites.All α n are thus purely offdiagonal, and encode the rung couplings of the Wilson ladder.In summary, the block entries α n and β n in Eq. (A33) have the structure, As argued with Eq. (A18), the coupling to the impurities is dominated at low energies by the symmetric channel.Therefore the coupling strengths along the symmetric and antisymmetric channel can differ, as quantified by the relative difference δ n .This splitting scales like δ n ∝ 1/x ≪ 1, and therefore is small for early Wilson shells close to the impurity for large impurity distance x ≫ 1.The case n = 0 is special, as it refers to the coupling of the impurities to the f0 states.In the spirit of symmetrically coupled Anderson impurities where there is actual hybridization of the impurities with the helical bath, it holds that δ 0 = 0 if x = x n is chosen on the discrete lattice Eq. (A11).In this case, the f0 states themselves are orthogonal already, and so there are strictly no impurity cross-coupling due to non-orthogonality of the f0 states. The fact that the block-tridiagonalization can proceed in real arithmetic has several advantages, in practice.The switch to the symmetric / antisymmetric basis prevents certain numerical errors from piling up during the blocktridiagonalization that are related to slowly drifting complex phases.Also by dealing with real arithmetic, phase conventions on basis states only refer to signs.The pair of states within every tridiagonalization step represent symmetric / antisymmetric states, and hence are already orthogonal, by construction, so there is no immediate explicit need for a Schmid decomposition within the pair of states fn . The total combined hybridization strength of the two impurities for given spin(-up) is [cf.Eq. (A15)] where the prefactor of 2 derives from the two impurity channels.This total combined hybridization strength is the same, on average, for the impurities f−1 in the symmetric / antisymmetric basis since l tr(T l T † l ) In the symmetric basis, however, the contribution of the individual channels can differ from each other, as reflected also in δ 0 ̸ = 0 symmetry, which is absent in a helical system.Rather, it is reduced to a discrete Z 2 particle-hole symmetry [cf.Sec.A 5].This is specific to the effective model of the helical edge used here, though, stressing that such a particle-hole symmetry is absent in 2D time-reversal invariant topological insulators [53].The complex phases with a 0 and b 1 close to the impurity are important and hence cannot be gauged away, since e.g.within a fixed spin flavor, the helical Hamiltonian is not time-reversal invariant [if derived from a real-space lattice the Hamiltonian necessarily would have to include complex spinorbit coupling terms; in the diagonal eigenbasis, as in Eq. ( 1), any quadratic Hamiltonian becomes real of course]. The coupling a 0 is fully determined by ⟨0\| f0+ Ĥbath f † 0− \|0⟩ with \|0⟩ the vacuum state, and hence can be expressed analytically, The precise value for a 0 is thus sensitive on the discretization, as seen with the second line.In the continuum limit, ε l → ε may be pulled inside the integral, which yields the last expression having used εe iτ ε = (−i d dτ )e iτ ε .While r 0 = 0 on the grid (A11), the derivative is generally non-zero with alternating signs and decaying like 1/τ ∼ 1/x.Therefore a 0 is non-zero for any x = x n .That is, the two impurities weakly see each other right away upon a single application of Ĥbath .On intuitive grounds, one can compare this to a tightbinding chain with long-range hoppings that decay like 1/x [cf.Eq. (A12)] which also gives rise to a similar behavior. d. Decoupling of anti-symmetric sector at low energies The antisymmetric sector gets suppressed once energies are much below ε ≪ E x ≡ v x [Eq.(A9)], or more precisely in the limit \|τ ∆ϵ l ≪ 1\|, This occurs in practice at very low energies in the NRG context.With r l → 1 − , while s l+ approaches the finite value of 2 − , the smaller eigenvalue s l− → 0 + approaches zero as 1 − 1 − which is numerically ill-conditioned.Hence form a numerical perspective, it is computed via an expansion around small ξ ≡ τ ∆ϵ l /2 = x∆k l /2, i.e., from Eq. (A17b), When setting s l− strictly to zero below some threshold s l− < 10 −16 , the block tridiagionalization eventually switches over to a hopping amplitude that decays twice as fast, because the antisymmetric levels that actually couple have been exhausted.On the other hand, keeping the asymptotic dependence in Eq. (A49) down to the lowest energies considered, the hopping amplitudes along the Wilson ladder always decay like ω n ∼ Λ −n/2 .That is, the antisymmetric channel x [Eq.(A9)] one may think of the bath states coupled to their respective impurity.Therefore the physics represents two independent copies of one-impurity problems down to energies Ex.Around the energy scale of Ex the impurities start to coherently interact with each other (if the impurities are, for example, already Kondo screened above the energy Ex, then the two impurities remain decoupled down to zero energy).The 2-impurity physics takes place at energies ε ≲ Ex where the relevant description of the bath effectively changes from the local (R, L) to the symmetrized (+, −) representation.The antisymmetric channel (η = −) starts to smoothly decouple, but stays in the system as a passive spectating bath space.The symmetric channel (η = +) is the one that remains fully coupled to the impurities.While the representation of the bath in the main text is always in the symmetric/antisymmetric configuration [cf.Eq. (A42), except for a final rotation of f0 back to the local representation].If the same unitary mixing of modes were to be applied for subsequent Wilson shells still, the property of having two independent copies of bath modes remains intact down to Ex. remains in the system, throughout.The reason for this is that the decoupling occurs smoothly.So once the energy scale (or more precisely, the energy resolution) drops below v/x, the antisymmetric channel does not decouple in an instant, and so it stays in the system, as schematically depicted in Fig. 9.At energies much below v/x, however, one can show in practice that the Wilson chain, indeed, switches over to two fully decoupled chains [cf.Fig. 10(b)].While the symmetric sector which remains coupled to the impurities, shows a smooth decay of the hopping amplitudes, in the antisymmetric channel the hopping amplitudes along its corresponding leg in the Wilson ladder becomes increasingly alternating (lower legs in Fig. 10): namely the paired up antisymmetric levels at energies ±ε l .They form strong bonds along the Wilson chain at zero energy, where bonding and antibonding states reveal the original ±ε l states in the star geometry. Since the impurity distance directly enters the coarse graining in the NRG, there will always be a qualitative change in the NRG energy flow diagram around the energy scale v/x [cf.Figs. 9 and 10].But this change in the representation of the bath can become irrelevant for static or dynamic properties from the perspective of the impurity.In this sense, the 'apparent' energy scale strongly visible in the standard NRG energy flow diagrams at the energy scale v/x may be irrelevant for the impurity.Nevertheless, this may leave minor artificial fea- .They are all real-valued, where black (red) color shows positive (negative) hopping amplitudes, respectively.Around the energy 1/x (shell n ∼ 40), the structure of the Wilson ladder changes, as qualitatively already argued in Fig. 9.The upper leg corresponds to the symmetric (η = +), and the lower leg to the antisymmetric channel (η = −).The rescaled Creutz couplings have amplitudes below double precision accuracy, hence are absent.(b) Same as in (a), except that starting from the position 'blocktrafo' towards the right a nearest-rung shell-mixing numerically determined block transformation was performed on top of (a).This shows that at low energies the system can be exponentially decoupled towards later shells into two independent channels.tures (wiggles) in the temperature or frequency dependence of physical properties around the energy scale of v/x.This effect is expected to be more pronounced for coarser discretization (larger Λ), but to diminish for smaller Λ. e. Block-tridiagonal structure for opposite spin Switching the spin has the same effect as changing the sign of energy or spatial inversion [cf.discussion after Eq. (A18)].This is also explicitly reflected in the the variable τ ≡ σx v [Eq.(A4a)] that appears in much of the above treatment.Therefore by construction of the starting vectors for the blocktridiagonalization of the spin-up channel above, from the point of view of the impurity, spin-down couples to an identical Wilson ladder of its own, except that the local f0 modes couple to the impurity levels in reverse order.In this sense, the backtransformation to the local representation of the impurity in Eqs.(A47) is useful as a prior step.Then the Pauli matrix σ x below accounts for the reversed order, while everything else remains the same for spin down as for spin up.The effect of the cross terms on the impurity spectral function, for example, are therefore, This makes intuitive sense, since motion in between the two impurities occurs in opposite directions for different spins. Dynamical correlation functions Within the NRG approach, correlation functions are computed by evaluating the Lehmann representation [33,49] which can be carried out exactly at arbitrary temperature T within the full density matrix approach (fdm-NRG [34]).To be specific, a retarded dynamical correlation function for two local impurity operators B and Ĉ, with ϑ(t) the Heaviside step function, B(t) = e i Ĥt Be −i Ĥt , and ⟨. ..⟩T the thermal average, is computed via its spectral Lehmann representation with a and b complete sets of many-body eigenstates, having Ĥ\|a⟩ = E a \|a⟩ with E ab ≡ E b −E a , thermal weights ρ a = 1 Z e −βEa , β ≡ 1 T , and Z ≡ a e −βEa the partition function.The iterative diagonalization within the NRG approach explicitly generates the full many-body state space Ĥ\|a⟩ ≃ E a \|a⟩ above.In practice, this yields a tractable symmetry-respecting, eigenstate decomposition of the entire impurity Hamiltonian on the Wilson chain [54] which while approximate, is well-controlled and complete. Standard fermionic and bosonic correlation functions have an (anti)commutator in their Green's function, with the commutator (s= − 1) for bosonic correlations such as spin-spin correlation functions, or the anticommutator (s= + 1) for fermionic correlation functions such as the fermionic local density of states.Equivalently, in frequency space G R BC (ω) = G BC (ω) + s G C † B † (−ω), where by construction at zero temperature A BC (ω < 0) = 0 since E ab ≥ 0 in Eq. (A53), such that the two contributions in the (anti)commutator to the full correlation function are separated in frequency space, since they exclusively contribute to positive or negative frequencies only. The important point with Eq. (A53), as already also pointed out with the hybridization function in Sec.A 1, is that for offdiagonal correlations B ̸ = Ĉ, the spectral data cannot only be negative, but fully complex, i.e., A BC (ω) ∈ C, if the matrix elements themselves are complex.In this sense, one cannot simply write the spectral data A(ω) as − 1 π Im G(ω).Instead, by only taking − 1 π Im (. ..) of the propagator 1 ω + −E ab , the full Green's function in frequency space can be simply obtained by standard means, i.e., via Kramers-Kronig transformation, or by folding with the propagator, By construction, with Eq. (A53) one still also has full access to the well-known simple spectral sum rules, which may be complex for B ̸ = C † in the present helical setting.This is relevant, for example, when computing the oneparticle correlation function across the impurity ⟨ dσL ∥ d † σR ⟩ ω [e.g., see Eqs. (A6) or (A7) for the non-interacting case]. Symmetries The global symmetries of the effective 1D model of the helical edge in Sec.II also manifest themselves in the structure of correlation functions.Aside from the symmetry U(1) charge ⊗ U(1) spin already discussed when introducing the helical model system in Sec.II and also explicitly exploited in our numerical simulations, the isolated helical edge can support further symmetries that are actually absent in the original full-fledged 2D topological system.The original topological aspect is reflected here in the fact that the isolated helical edge exists as a valid physical model in the first place.With this in mind, the isolated helical edge with two Kondo impurities (2HKM) located symmetrically around the origin also preserves • Z 2 time reversal symmetry (Z TRS ): momentum together with spin reversal is preserved by the helical edge.The local impurity-bath Kondo interaction is also spin-reversal symmetric, with the impurities themselves located symmetrically around the origin [cf.Eq. ( 5)].Therefore k → −k together with the reversal of the impurities, η → −η, leaves the local impurity-bath interaction invariant.Now the combined operation k → −k and η → −η is equivalent to spatial inversion.Hence, for our model setup with an isolated helical edge, TRS can be translated into spatial inversion with simultaneous global spin reversal. • Z 2 particle-hole symmetry (Z p/h 2 ): the helical channel in Fig. 1 was chosen such that for every level at ε kσ > 0 there is a level at ε −kσ < 0, having ε kσ = −ε −kσ .By having half-filling, this converts a particle to a hole or vice versa.Furthermore, by construction, the Kondo interaction of the impurities with the helical channel are also particle-hole symmetric. The Hamiltonian in Sec.II has two impurities located symmetrically around the origin [cf.Eq. ( 5)].From an Anderson impurity point of view with explicit hybridization as in Eq. (A2), this is the only point where complex numbers enter the total Hamiltonian.From a numerical perspective, the Fermionic operators ĉkσ and dησ can be encoded by real matrix elements, such that the only complex entry is the phase e ikx η 2 .This also holds when switching from Anderson-type hybridization to Kondo spin interactions when represented in terms of Ŝη ± or Ŝη z , as their matrix elements are also real.With this perspective, it holds that complex conjugation of the Hamiltonian, Ĥ → Ĥ * , is equivalent to reversing the locations of the impurities dησ → d−η,σ .Denoting the latter by R I , with R I H\|a⟩ = H * R I \|a⟩ it holds that if \|a⟩ is an eigenstate of H to eigenvalue E a , then so is \|a ′ ⟩ ≡ [R I \|a⟩] * .Hence, with K denoting complex conjugation, R I ≡ R I K is an anti-unitary symmetry of the system, with \|a ′ ⟩ = R I \|a⟩ also an eigenstate of the Hamiltonian with the same eigenenergy. As a consequence of the above symmetries, for example, the spectral data of spin-spin correlations as in Eq. (A56) is real, after all.Based on the Lehmann representation in Eq. (A53), one encounters the matrix elements For the case that η = η ′ , i.e., intra-impurity spin correlations, this product of matrix elements can be combined into \|⟨a\|..\|b⟩\| 2 , which is real.For the case of inter-impurity spin correlations, η = −η ′ , taking the complex conjugate and inserting Now since a and b can be chosen to also be simultaneous eigenstates of R I , i.e., having a = a ′ and b = b ′ with the same eigenvalue w.r.t.R I , the Lehmann sum of the spin-spin spectral functions, while having complex matrix elements, yields a purely real result.In practice, a and b are not eigenstates of R I .Yet by explicitly resorting to complete many-body basis sets within the NRG [54] evaluated within fmd-NRG [34], the spectral data has an imaginary contribution with relative strength comparable to numerical noise based on double precision accuracy, and hence can be ignored. From Anderson to (anisotropic) Kondo type model The entire discussion above assumed Anderson-type impurities that hybridize with the helical edge mode.Now if the local Coulomb interaction U with each impurity is large, charge fluctuations get frozen out.Therefore in the lowenergy regime, charge fluctuations can be integrated out locally with each impurity via Schrieffer Wolff transformation.Assuming half-filling of each impurity with a single magnetic spin-half moment, second order perturbation theory based on Eq. (A2) yields the Kondo-type interaction f † 0ησ f0ησ ′ the spin operator with respect to the bath site at the location of impurity η.Therefore a single impurity interacting with in a helical edge mode is identical to a regular 1-impurity Anderson or Kondo model without a helical character.The dynamically generated low-energy Kondo scale 47] for the 1-impurity problem with ρ 0 the one-particle density of states around the Fermi level, however, also represents a relevant energy scale for the 2-impurity problem.Assuming the impurity distance on the grid (A11), the local bath operators f0ησ are already properly orthonormalized, such that there are no issues with crosstalk between the impurities. For spin-independent hybridization between the Anderson impurities and the bath, the resulting Kondo coupling in Eq. (A59) is SU(2) spin symmetric.While bearing in mind that the helical bath mode itself already has the SU(2) symmetry broken, the spin symmetry can also be broken at the level of the Kondo Hamiltonian, giving rise to anisotropic local spin-spin interactions.Assuming a single preferred direction (z) with J x = J y ≡ J ̸ = J z , J z > J (J z < J describes an easy-axis (easy-plane) regime, respectively.The global U(1) spin symmetry thus remains preserved. Effects of finite bandwidth with Kondo interaction When starting from the Anderson model, one may scale the local Coulomb interactions properly to infinity in relation to other parameters, with the result that also in the numerical setting, one effectively arrives at the Kondo model [55].For the Anderson model the effective bandwidth relevant for the impurities is cut off by U if U < D. However, by taking U ≫ D when transitioning towards the Kondo model, bandwidth keeps playing a considerable role, and the universal wide-band limit is approached rather slowly. Here for the anisotropic 2HKM in the present case, if the Kondo scale T K (J, J z ) ≪ E R ≪ D = 1 for each impurity in-dividually is just several orders of magnitude smaller than all other energy scales, effects of finite-bandwidth are still considerable.In order to reach the wide-band scaling limit (here J ≲ 0.02), the single-impurity Kondo scale is already many many orders of magnitude lower than the band width D = 1 (T K ≲ 10 −40 ), as demonstrated in Fig. 11, and also consistent with the literature on the single-impurity case [55].In the present case, we see that similar arguments also carry over to the RKKY regime, even if Kondo physics is irrelevant (in the sense that it sets in at much smaller energy scales, even much smaller than any temperature of practical interest). In the RKKY regime and in the wide-band limit, the impurities are expected to be well decoupled from the bath and described by In practice, however, one sees substantial deviations from this expectation value up to nearly 20% for J = 0.1 in the NRG data even using J z = 0 as shown in Fig. 11(a).Hence these deviations must derive from higher-order processes that go beyond second order PT [see App.D].The coupling to the bath remains finite in the low-energy regime, thus inducing fluctuations in the impurity spin configuration.These deviations can be reduced systematically by lowering the Kondo coupling J, e.g., having a deviation already below 1% for J ≲ 0.02.As shown in Fig. 11(a) the static expectation value ⟨S z L S z R ⟩ approaches −1/4 in a polynomial fashion as J is lowered, down to the smallest J considered.Therefore, indeed, the deviations seen in Fig. 11(a) are clearly due to finite bandwidth.This demonstrates that finite bandwidth does play an observable role in the Kondo setting [55] when comparing numerical to analytical results if the latter strictly assumed the wideband limit.If one considers the wide-band scaling limit reached within deviations in observables of about 10 −3 , this suggests approximately J ≲ 0.02 in Fig. 11(a).On the single impurity level, this already corresponds to astronomically small Kondo scales T K ≪ 10 −40 in Fig. 11(b) consistent with earlier NRG studies [55].From a physics point of view, however, we do not expect that the observed minor variations change the overall physical picture in any significant qualitative manner.It follows from the above that where ±a 2 with a ≥ 0 is some constant of integration.If the starting point has \|z 0 \| > \|x 0 \| (easy-axis), then the positive sign is chosen for a 2 , whereas the regime \|z 0 \| < \|x 0 \| (easy-plane) has the negative sign.The contours described by Eq. (B3) exactly reflect the RG paths of the anisotropic Kondo (parabolas or hyperbolas separated by \|x 0 \| = \|z 0 \|, as shown by the gray lines in Fig. 12).It simply also follows from Eq. (B2) that for x → 0, (x, z) stops flowing.The model is physically equivalent for x 0 → −x 0 (vertical flip in Fig. 12) as this can be absorbed into a gauge transformation of the spin basis. For z 0 > 0 and \|x 0 \| > \|z 0 \|, the anisotropic Kondo model flows to the isotropic strong coupling regime.That is integrating out the bath starting from large D needs to be stopped at some D * > 0 where x and z diverge, which thus defines the dynamically generated low-energy Kondo scale T K := D * . In the easy-axis Kondo regime (J z > \|J x \|), with z ≥ a, the RG differential equations diverge at The Kondo scale vanishes to exponentially small energy scales for J x → 0, i.e., a → z − 0 .In the easy-plane regime (\|J x \| > \|J z \|), with \|x\| ≥ a, the RG differential equations diverge at The single-impurity Kondo scale from the previous section may be compared to the RKKY energy E R [cf.Eq. ( 20b)].The latter decays with distance, hence is largest for short distances, with the shortest possible distance given by one 'lattice spacing' [cf.Eq. ( 2)], hence E R ≤ (ρ 0 J) 2 in the adopted units [Sec.II A].With this in mind, Fig. 12 also includes contours of constant T K (J, J z ) − (ρ 0 J) 2 ≤ 0 (green lines), with the thick line representing T K (J, J z ) = (ρ 0 J) 2 .Therefore the parameter regime to the right of the thick green contour always has T K > E R for any x ≥ 1, meaning that in this regime Kondo is always dominant over RKKY.This regime has been predicted in Ref. [1].In particular, this includes the isotropic Kondo for J ≳ 0.25, corresponding to a dimensionless coupling j 0 ≡ ρ 0 J = 0.125.While still clearly below 1., this is on the upper end of what may be considered acceptable on physical grounds for the Kondo model: Note that the poor-man's scaling approach for the Kondo model only includes second order virtual processes by integrating out the bath which assumes (and thus is justified only if) J 2 D ≪ D, i.e., j 0 ≪ 0.5.Conversely, note that coming from an Anderson model, one has j eff 0 ≃ 4Γ πU with hybridization strength Γ and onsite interaction U .So in order for the Kondo model to be justified in the first place, the onsite interaction U needs to be large enough so that charge fluctuations can be integrated out via virtual second order processes (cotunneling) in order to obtain a pure effective spin Hamiltonian.This clearly requires U ≫ Γ, and thus also j 0 ≪ 1 on physical grounds. Conversely, to the left of the thick green line in Fig. 12, RKKY can dominate in the low-energy regime if a pair of impurities is just brought close enough to each other.In particular, RKKY also occurs within the entire white region to the left where T K = 0.With E R > T K = 0 then, the impurity distance x can be taken to infinity while still seeing RKKY in the low-energy regime (note that since RKKY is second order, with E R ∝ J 2 /x, the sign of the Kondo couplings is irrelevant). Appendix D: Second order perturbation processes entering RKKY Complimentary to the field theoretic approach in the main text, the RKKY processes can also be analyzed from the point of view second order perturbation theory (PT).Ultimately, this generates the same effective Hamiltonian as in the field theoretic approach, but purely within the Hamiltonian setting, e.g.via the Feshbach formalism [58,59]. Let the full Hamiltonian of two Kondo-coupled impurity spins η ∈ {R, L} ≡ {+1, −1} at location ηx 2 , i.e., at mutual distance x, interacting with a one-dimensional helical bath encoded by the Fermionic operators ĉkσ be given by with a ∈ x, y, z, and bath spins (with τ a the Pauli matrices), assuming a total of N levels k for each spin.Furthermore, the energies of the bath ε kσ = σvk are constrained to within a finite half-bandwidth \|ε kσ \| ≤ D. Taking J≡J x =J y and σ± η ≡ ŝ± η ≡ ŝx η ± iŝ y η , the Kondo interaction can be written from the point of view of the helical Fermions as, with the matrix notation in σ∈{↑, ↓}≡{1, 2} for the operators of each impurity [see also Eq. ( 13)], with anisotropy ∆≡ Jz J and τ ± = 1 2 (τ x ± iτ y ).The isotropic case ∆=1 just reduces to T η = Ŝ † η • τ .The effective Hamiltonian in between the impurities, i.e., their direct interaction, is generated via second order PT in the Kondo setting (4 th order in the hybridization for an Anderson model).Then with the shortcuts 1 ≡ (k 1 σ 1 ), etc., a typical second order process is given by with degrees of freedom such as impurity location η∈{L, R}≡{−1, 1}, or spin σ or momenta k to be summed over eventually.P 0 is a projector into the target low-energy regime of the bath without acting at the impurities.This is in the spirit of generating the low-energy effective Hamiltonian such as the Feshbach-Fano partitioning [58,59].There, by construction, one needs to start out of the low-energy regime, say by considering some infrared cutoff energy D * ≪ D. In the low-energy regime described by P 0 then, all bath levels with energy ε kσ < −D * are strictly occupied and all energy levels ε kσ > +D * are strictly empty.Now in Eq. (D2), the perturbation T on the r.h.s.generates a particle-hole (p/h) pair with particle energy ε 3 > 0 and hole energy ε 4 < 0. There-fore as long as either ε 3 > +D * or ε 4 < −D * (or both), this represents state space outside the low-energy regime P 0 .If D * ≪ D, the overwhelming number of processes will involve both, particle and hole at energy cost above D * .There in order to return to the low energy regime, exactly the same p/h pair must be destroyed again.This constrains the second processes above to δ 14 δ 23 , with the shortcut notation δ ij ≡ δ kikj δ σiσj .Actually, the processes δ 14 δ 23 can be included all the way down to zero energy, i.e., there is apriori no need for an infrared energy cutoff D * , so one can assume D * → 0 + as long as the sum is well-defined (which it turns out it is).With the bath projected via P onto the exactly filled helical Fermi sea, this generates the effective, purely local inter-impurity Hamiltonian (D3) having taken the continuum limit i ≡ (k i σ i ) → (ε i σ i ), with ρ 0 the constant one-particle density of states of the helical edge for a given spin at the Fermi level as in Eq. ( 2).The phase factor in the second line rewritten in terms of energies is already also specific to the helical edge.As an aside, we note that the energy denominator of I ηη ′ σ1σ2 can be rewritten into an integral over imaginary frequencies, which resembles a Matsubara summation at zero temperature.Via contour integral, the result is non-zero only if ε 1 and ε 2 have opposite sign as encoded into the Heaviside step function on the r.h.s.This already reflects the particle-hole nature of the excitations in Eq. (D3) where ε 1 < 0 and ε 2 > 0. The case with opposite signs is included in Eq. (D3) via the overall sum.Assuming η̸ =η ′ , then the case η ↔ η ′ gives the additional contribution above, while also exchanging the labels 1 ↔ 2, such that the sign in the phase factor is properly restored, while also having [ T η , T η ′ ]=0 for η̸ =η ′ .Hence by summing over all second order processes, one can connect the second order perturbative approach here to the doublepropagator structure in Eq. ( 15) in the main text obtained from the field-theoretic approach.Collecting phase factors in k 1 and k 2 , the latter thus permits the interpretation that any second order process in the effective impurity Hamiltonian requires the free propagator of two particles shuttling back and forth in between the impurities (one particle needs to propagate the distance +x, and another the distance −x).Here in the helical setting, however, the directions that particles can move are constrained depending on their spin.This manifests itself in the overall structure of the resulting RKKY Hamiltonian [1].The integral I ηη ′ σ1σ2 in Eq. (D3) is generally well defined when both, ε 1 and ε 2 approach zero energy, irrespective of the phase factor since The imaginary part form i0 + does not contribute, as by the integral limits, it can only contribute at ε = ε ′ = 0, where by the double integral the integrated weight vanishes.When η=η ′ , the complex phases drop out, and with S 2 η ∝ 1 1 the integral in Eq. (D3) just generates a plain irrelevant shift in the global energy reference as estimated above (the generation of the single impurity Kondo couplings needs to consider finite D * and relax the condition δ 14 δ 23 above).Hence the following discussion focusses on η̸ =η ′ , i.e., η ′ = − η, in which case the phase factors in the enumerator are non-trivial and reflect the underlying helical physics.In contrast to the energy denominator, however, the phase factors carry momenta as arguments.Converting these into energies, in the helical setting with ε kσ = σvk, this involves signs depending on the spin as already indicated with the integral I ηη ′ σ1σ2 in Eq. (D3).Contributions that involve a spin flip (σ 2 = − σ 1 ), will have opposite relative sign of ε 1 vs. ε 2 in the phase factor in Eq. (D3) as compared to the energy denominator, whereas in the absence of a spin flip the relative sign is the same.To evaluate this integral including the phases, it is therefore convenient to change variables to symmetric and antisymmetric combinations With ε 1 < 0 and ε 2 > 0, by construction, 0 ≤ ε ≤ 2D and ε ′ ∈ [−ε, ε].The integral will converge with large D, such that the upper integral limit can be taken more loosely by deforming the integration area to ε ≲ D = 3D 2 , with the upper integral limit D assumed large, and eventually taken to infinity if well-defined.With this one obtains for the case including a spin flip, with τ ≡ with the dimensionless Kondo coupling strength j 0 .Overall, this generated direct impurity interaction is ferromagnetic and non-oscillatory with a plain decay with inverse distance x, having the RKKY energy scale as already encountered in Eq. (20) in the main text.The RKKY Hamiltonian is independent of the bandwidth D, and aside from the dimensionless Kondo coupling strength j 0 , only references the coherence scale E x ≡ 1 τ with τ = x/v the time required for a helical particle to travel from one impurity to the other.With the lattice spacing in Eq. ( 2), nevertheless, this energy scale may be rewritten as E x = D/π x/a .That is when measuring distance on the grid (A11), by definition, this involves a finite bandwidth, such that the bandwidth does appear in the definition of E x .In the continuum wide-band limit, however, the natural way to think about the coherence scale is E x = 1/τ without any reference to bandwidth. In the absence of a spin flip, i.e., σ 1 = σ 2 the integral can be similarly evaluated, 2 , assuming that the integral is well-defined in the wideband-limit.Again the contribution from the imaginary part i0 + in the denominator vanishes, since dε δ(ε) εe i τ ε → 0, as already expected from Eq. (D5).By summing over spin or location, with τ ∝ ησ 2 , only the real part remains, ησ1 I η=−η ′ σ1=σ2 ≃ 4 τ sin τ D which seems to suggest 1/x behavior similar in magnitude to Eq. (D8).In the present case, however, the integral remains sensitive on the bandwidth D. Therefore the assumption for introducting D above does not hold, and the integral needs As apparent from the oscillating averaging structure in the second term, it also vanishes for large λ.In the asymptotic form for large λ, the leading term of the cosine integral Ci(λ) ∼ sin λ λ again drops out on the grid (A11).Therefore the subleading term Ci(λ) ∼ cos λ λ 2 becomes the dominant one for large λ.But with DCi(λ) ∼ D λ 2 ∼ 1 Dx 2 , this does not just decay faster over distance as compared to RKKY for a normal 1D metallic mode, but is also suppressed in the wideband limit.Therefore as expected the ZZ contribution to the RKKY Hamiltonian properly vanishes for D → ∞ even for finite J z .At finite bandwidth, however, there is a finite return probability, resulting in a small but finite S z L S z R interaction strength across the impurities.This may be considered acceptable on physical grounds, bearing in mind that for a true 2D model the spin-dependent directedness of motion only concerns the edge but not the bulk states gapped out to energies > D. Finally, we point out that above integrals also appear in the theory of a normal metallic edge when one assumes the same linear dispersion and finite bandwidth as for the helical system here.However, by having additional branches of the opposite helicity in the dispersion for back-propagation, the same integrals appear and are summed over all spin interactions XX, YY, and ZZ.This way the RKKY interaction becomes isotropic for a normal metallic edge.The above argument that the leading oscillatory term vanishes is particular to the one-particle dispersion chosen here, and would also apply to the normal metallic edge with the dispersion indicated.This differs crucially from a system of free particles in a Fermi sea with a quadratic dispersion and a finite Fermi energy.In this case, the analytic structure of the respective integrals is different.As a consequence, this permits the leading 2k f oscillatory term ∝ 1/x to be present in a normal metallic 1D mode [43]. FIG. 3 . FIG.3.NRG analysis of the 2HKM for J = 0.1 (i.e., j0 = 0.05) -Each row corresponds to a different integer impurity distance x as indicated with the left panel, which increases exponentially from top to bottom.The vertical axis is the same for all panels, namely Jz/J ≡ jz/j0, whereas the horizontal axis represents energy in various forms.Here we adopt the NRG energy scale point of view, throughout, which starts at large energies at the left and then proceeds towards exponentially smaller energy scales towards the right.Rows (columns) are labelled by letters (numbers), respectively, e.g., having (a1) for the upper left panel.The vertical black dotted [solid blue] marker replicated in all panels indicate the coherence scale Ex (33) [RKKY scale ER (20)], respectively, as labelled in panel (a1), having Ex/ER = 1/πj 2 0 = 127.3,throughout.Similarly, the brown solid curved line shows the analytical single-impurity Kondo temperature TK(J, Jz) [App.B] for reference.This curve TK is visually cutoff at ER, because TK is irrelevant at lower temperatures.The brown marker in panel (h2) shows the finite intercept TK(J = 0.1, Jz) = 4.31•10 −8 at Jz = 0.Each column shows the quantity indicated above the top panel: Panels (:,1) [first column] show the static (equal-time) inter-impurity correlations ⟨S L z S R z ⟩ vs. temperature and Jz.Panels (:,2) [second column] gives a visual impression of the changes along the NRG energy flow diagram vs. energy scale ωn ∼ Λ −n/2(for precise prefactors, see[46]).This encodes cumulative effective thermal weights in three arbitrary but fixed low-energy symmetry sectors into red-green-blue (RGB) colors for all odd Wilson shells n (to avoid even/odd effects) based on an effective inverse temperature βn ≡ 4ωn.The symmetry sectors chosen for this were q ≡ (Q; S tot z ) ∈ (0; 0, 1, −1) with Q the total charge relative to half-filling.The remainder of the columns shows dynamical spin-spin correlation functions ⟨ Ŝη ∥ Ŝη ′ † ⟩ω as indicated at the top of each column at zero temperature (T = 1.8 10 −11 ).The solid-dotted lines shows the respective derived inverse static susceptibility T ηη ′ FIG. 4 . FIG. 4. NRG spectral data of the 2HKM for J = 0.1, Jz = 0, x = 10 for the dynamical correlations functions as partly indicated with the left axis [transverse in the upper panels, longitudinal in the lower panels; intra-impurity for the left panels, whereas in the right panels also includes the inter-impurity correlation for comparison; note the sign in the legend with panel (d)] -Left panels: individual spectral data for z-shifted logarithmic discretization (nz = 8 curves with z ∈ [0, 1[, having Λ = 4, broadening σ = 0.1), and the corresponding z-averaged data in the right panel.The horizontal and vertical axis are globally scaled by the z-averaged low-energy scale TS z = 1.4•10 −4 = 0.556 ER only, so with spectral sum rules in mind, the data shown is of order one. FIG. 8 . FIG.8.Low-energy scales vs. distance from inverse static susceptibilities [cf.Eq. (34)] as indicated in the legend of panel (a) for Jz = 0, throughout.Left panels (J=0.05)[right panels (J=0.10)]have temperature T < 10 −10 [the Kondo temperature TK ∼ 4•10 −8 ] as low-energy cutoff, respectively.The blue vertical markers translate the low-energy scale to distance, i.e., x0 ≡ j 2 0 /max(T, TK) which also corresponds to the crossing point of ER (black dashed) with max(T, TK) (horizontal marker).The lower panels redraw the data in the upper panels, but vertically scaled by TS.The shading indicates the standard deviation of the color matched energy scale due to z shifts in the NRG discretization. FIG. 9 . FIG. 9. Schematic representation of the 2-impurity Wilson setup of the bath giving rise to two intercoupled chains (for a quantitative example, see Fig.10).The system moves to small energies towards the right with the Wilson shells n having energy εn ∼ Λ −n/2 .At high energies ε ≫ Ex ≡ vx [Eq.(A9)] one may think of the bath states coupled to their respective impurity.Therefore the physics represents two independent copies of one-impurity problems down to energies Ex.Around the energy scale of Ex the impurities start to coherently interact with each other (if the impurities are, for example, already Kondo screened above the energy Ex, then the two impurities remain decoupled down to zero energy).The 2-impurity physics takes place at energies ε ≲ Ex where the relevant description of the bath effectively changes from the local (R, L) to the symmetrized (+, −) representation.The antisymmetric channel (η = −) starts to smoothly decouple, but stays in the system as a passive spectating bath space.The symmetric channel (η = +) is the one that remains fully coupled to the impurities.While the representation of the bath in the main text is always in the symmetric/antisymmetric configuration [cf.Eq. (A42), except for a final rotation of f0 back to the local representation].If the same unitary mixing of modes were to be applied for subsequent Wilson shells still, the property of having two independent copies of bath modes remains intact down to Ex. FIG. 10 . FIG.10.Typical Wilson ladder for the 2-impurity helical system as described by Eq. (A33) for an impurity distance x = 10 6 .The labels on top indicate the Wilson shell n (based on Λ = 2).The widths of the bonds are proportional to the hopping amplitudes rescaled by Λ −n/2 .They are all real-valued, where black (red) color shows positive (negative) hopping amplitudes, respectively.Around the energy 1/x (shell n ∼ 40), the structure of the Wilson ladder changes, as qualitatively already argued in Fig.9.The upper leg corresponds to the symmetric (η = +), and the lower leg to the antisymmetric channel (η = −).The rescaled Creutz couplings have amplitudes below double precision accuracy, hence are absent.(b) Same as in (a), except that starting from the position 'blocktrafo' towards the right a nearest-rung shell-mixing numerically determined block transformation was performed on top of (a).This shows that at low energies the system can be exponentially decoupled towards later shells into two independent channels. 2 A58) which is diagonal in η.When projecting the into the lowenergy Kondo regime of the Anderson model by scaling up local interactions, this leaves the representation of the bath untouched.Therefore from the NRG point of view the bath remains completely unaffected by whether one resorts to an Anderson-type or low-energy Kondo-type impurity setup.Based on the coarse-grained version in Eq. (A32) then, ĤK ∼ 2J η∈L,R Ŝη • Ŝ0η , (A59) with J the Kondo coupling, Ŝη ≡ Ŝdη the spin operator of impurity η, and Ŝ0η ≡ σσ ′ τ σσ ′ 2 FIG. 11 . 2 x FIG. 11.Static ⟨S z L S z R ⟩ impurity correlations for TK ≪ ER ∼ J 2 x obtained by NRG towards the wide-band limit D ≫ J(= Jx = Jy) for Jz = 0. (a) Deviations of the computed ⟨S z L S z R ⟩ from the expected RKKY value of −1/4 on a loglog plot.Data is shown for various impurity distances x, with the polynomial fit obtained for x = 10 showing approximate quadratic behavior (red dashed line).(b) Comparison of energy scales for the same parameter range as in (a) where the temperature T ∼ 10 −9 used in NRG simulations in (a) is indicated by the horizontal gray line.The data for the Kondo scale TK was obtained via poor-man's scaling [cf.App.B].Red line shows a simple estimate for the RKKY energy at x = 10, demonstrating that (a) is deeply in the RKKY regime, i.e., ER ≫ TK, T . FIG. 12 . FIG. 12. Kondo scale of the anisotropic Kondo model from poor Man's scaling [Eqs.(B4) and (B5)].Jx < 0 is physically equivalent to Jx > 0, such that the lower half of the panel is a mirror image of the upper half.The thick black horizontal line (Jz < 0 for Jx = 0) represents a stable line of fixed points without a Kondo scale.The entire white region below the diagonal lines to the left flows towards it.The black dotted horizontal line (Jz > 0 for Jx = 0) represents a line of unstable fixed points that flow to strong coupling for any small but finite Jx (strictly at Jx = 0 one has TK = 0, represented by white color).The gray lines represent RG flow contours as in Eq. (B3).The Kondo temperature is defined by the starting point (x0, z0) on such a contour.The Kondo scale increases monotonically from left to right, and also with increasing \|Jx\|.The green contours show lines of constant TK (J, Jz) − (ρ0J) 2 ≤ 0. The thick green contour to the right describes TK (J, Jz) = (ρ0J) 2 = ER(x = 1) which represents the largest RKKY energy possible by putting two impurities right next to each other, with the closest distance being one 'lattice spacing' [cf.Eq. (20b)]. with to be evaluated more carefully.The exact representation of the integral in the first line of Eq. (D10) yields with z ≡ τ ε, λ ≡ τ D, and λ ≡ \| λ\| = τ D, Ci() the cosine integral function.The first term vanishes on the grid (A11) having \|λ\| = τ D = πx/a a multiple of π.	2022-08-05T06:41:37.055Z	2022-08-03T00:00:00.000	{ "year": 2022, "sha1": "ecb81312ff678e07da15208ddc8a417233e2439c", "oa_license": "CCBY", "oa_url": "http://link.aps.org/pdf/10.1103/PhysRevResearch.5.033016", "oa_status": "GOLD", "pdf_src": "ArXiv", "pdf_hash": "ecb81312ff678e07da15208ddc8a417233e2439c", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
258663274	pes2o/s2orc	v3-fos-license	Perceptions of Students About the Use of Webinars in Classrooms: A Case of Abu Dhabi University This paper investigates the use of webinars among students at a university in Abu Dhabi using UTAUT model. Cross-sectional research design was used, and 134 participants were involved through a purposive criterion sampling technique. Data were collected through a survey based on a questionnaire. Findings revealed a significant relationship among performance expectancy (p-value 0.000), effort expectancy (p-value 0.000), and social influence (p-value 0.000), and an insignificant relationship with facilitating conditions (p-value 0.10) and voluntariness of use (p-value 0.61). The results would facilitate educational institutions in implementing the advanced system of education incorporating the use of webinars as a contemporary technological tool. The study is significant in determining the need for the use of webinars in class, illustrating both students’ and teachers’ perspectives. The implementation of the proposed results is not only limited to the case of Abu Dhabi but can be generalized to other educational institutions as well. INTRoDUCTIoN Technology use in education has gained significant attention, as it enables the development of advanced educational institutions that incorporate emerging methods of teaching and learning processes (Yanuschik et al., 2015).Students emphasize the need for technology, such as tablets and cell phones, to access e-books (Vassilakaki et al., 2016).While technological devices may reduce the teachers' efforts to assess student performance, they also offer opportunities for online testing and web-based quizzes in the modern age (Chiu & Churchill, 2016;Tsinonis, 2018).Implementing modern technology in various universities in Abu Dhabi has reduced staff time and effort (Al-Qirim et al., 2018).Studentcentered approaches in modern universities focus on providing easily accessible, relevant content to students, which develops learners' autonomy and prepares them for modern challenges (Abdallah, 2018).In Abu Dhabi institutions, nearly 40% of teachers use modern technology for instructional, administrative, and record-keeping purposes (Abdallah, 2018). The significant use of technology highlights how it is crucial for both teachers and students, as a majority of tasks related to teaching and learning are based on technology use (Al Musawi et al., 2016).Moreover, the increased use of modern technology in Abu Dhabi's educational institutions has had a positive impact on the state's educational development.The growth of computer software and hardware has enabled students to integrate technology into their pursuit of quality education, thus opening up new pathways for higher education by replacing previously ineffective teaching and learning methodologies (Raji, 2019).According to Musah et al. (2023), the concept of quality should be perceived as a continuous process of advancement instead of a fixed destination. Using technology in higher education is vital, according to Arkorful and Abaidoo (2015), as it allows us to modernize the education system with a balanced use of technology.This calls for new, innovative teaching methods to transfer information to society.However, some teachers and members of the administration feel that we must complement modern technology introductions with clear ICT instructions, since many aged teachers lack familiarity with modern technology (Ruggiero & Mong, 2015).Therefore, many educators are now required to plan their courses by integrating the use of technology (Islim & Sevim Cirak, 2017;Liu et al., 2015).Various tutorials, online materials, videos, and animations are now included as part of modern education processes (Instefjord & Munthe, 2017).Using technology in classrooms provides visualization of various concepts and ideas, making it easier for students to comprehend course topics (Abdallah, 2018).Additionally, it promotes self-efficacy among students, instilling confidence in them to handle everyday tasks efficiently.Sun et al. (2018) demonstrated that individuals with high levels of self-efficacy have greater abilities to deal with challenging situations that others may find difficult to handle. Various authors have studied the use and intention to use technology in the academic sphere (Khechine et al., 2013(Khechine et al., , 2014;;Lakhal et al., 2013), while others have investigated the impact of technology on student outcomes, such as learning (de Gara & Boora, 2006;Myers & Schiltz, 2012;Wang & Hsu, 2008), academic performance, and satisfaction (Khechine & Lakhal, 2015;Lakhal et al. 2014).Communication technology has become an integral part of knowledge exchange, and various advancements have made it a crucial tool for sharing and imparting knowledge.One such tool is webinars, which have an enormous potential that must be explored and investigated in the digital ecology of knowledge transfer (Gupta & Sengupta, 2021). Few studies have investigated technological behavior among Abu Dhabi students, and those that have done so did not focus specifically on a particular tool such as webinars.Given the integration of technology and the use of a holistic technological adaptation model, analyzing the determinants of technology use among students can provide a more comprehensive view of technology acceptance, usage in the classroom, and its consequences.Moreover, this study fulfills the recommendation of Tate et al. (2015) to address the research gap where there is vast research on TAM (Technology acceptance model) but scarce research on UTAUT (Unified theory of use and acceptance of technology). Using digital technology has led to the emergence of online learning as a promising approach to facilitate self-directed learning for students (Jandigulov & Tlepbai, 2018;Jandigulov, et al., 2023).This innovative technology employs a range of tools, including web servers, web pages and websites, email, chat rooms, videoconferencing, virtual classrooms, and wikis, to enable communication between teachers and students at different times and locations (Bush et al., 2019;Liu et al., 2018;Jandigulov et al., 2023). Moreover, the rationale for selection of a university in Abu Dhabi studies is based on the increased use of technology in the business sector and its recognition as the model nation for technology adoption (Cabral, 2018;Sadaqat, 2019).Therefore, this study investigates students' perception towards webinars and factors affecting it.The study is significant in determining the need for the use of Webinars in class, illustrating both students' and teachers' perspective.The objective is to analyze the role of Webinars in classrooms while providing valuable concern if is needed to be determined regarding its use.Policymakers in the educational sector can benefit from the findings of this study to demonstrate high-level regulations that produce efficient achievements of the modern education system.Implementing the proposed results is not only limited to the case of Abu Dhabi but can be generalized to other educational institutions as well. LITERATURE REVIEW Using technology in the educational sector has been gaining significant attention over the past few years.With the recent pandemic, integrating technology in education has become even more crucial.Among the various technological tools, webinars have emerged as a viable option for educators to conduct online classes and enhance the teaching and learning process.We expect the global pandemic to cause significant, on-going disruptions to the education sector, with the most apparent and pressing effect being the implementation of mandatory physical distancing measures (Abdallah & Riyami, 2022). This study aims to investigate the factors that encourage the use of webinars among students at a university in Abu Dhabi, using the Unified Theory of Acceptance and Use of Technology (UTAUT) model.We will highlight the UTAUT model as the most effective predictive model concerning the adaptation of technology and utilize it to gain a comprehensive insight into the user's behavior prediction. Theoretical Framework: UTAUT Model Various studies have emphasized the understanding of individual acceptance and use of technology (Benbasat & Barki, 2007;Venkatesh et al. 2003).Samaradiwakara and Gunawardena (2014) cite the work of Oye et al. (2014) showed that technology provides no value unless it is used or accepted.This makes it essential to investigate the factors which encourage the use of technology and determine the impeding blocks for the university students in Abu Dhabi.The unified theory of acceptance and use of technology (UTAUT) was developed by Venkatesh et al. (2003) and is seen as the most effective predictive model concerning technology adoption (Al-Shafi & Weerakkody, 2010;Alawadhi & Morris, 2008).We observe it as the synthesis of the previous eight technological acceptance theories; namely, theory of reasoned action (Fishbein & Ajzen, 1975), theory of technology acceptance model (Davis, 1989), theory of motivational model (Davis et al., 1992), theory of planned behavior (Ajzen, 1991), combined theory of planned behavior/technology acceptance model (Taylor & Todd, 1995), model of personal computer utilization (Thompson et al., 1991), diffusion of innovation theory (Rogers, 1995), and social cognitive theory (Compeau & Higgins, 1995).Khechine et al. (2016) showed how UTAUT helps develop a comprehensive insight into the user's behavior prediction, which individual modes could not attain.It consists of six primary variables, such as independent variable inclusive of effort expectancy, performance expectancy, social influence, and facilitation conditions, while the dependent variables include the usage behavior and the behavioral intention.While the model was primarily used for identifying employee technology acceptance, it has been gradually used in different contexts such as education and healthcare.For the present study, we used this model in the blended learning environment.A study conducted by Lieser et al. (2018) showed a process that comprised three stages for developing a webinar integration tool to increase the interaction between teacher and student and learning in blended environments.Similarly, Kohorst and Cox (2007) represented that a webinar is very operative.Moreover, several other authors also discussed the various benefits of using webinars in classrooms (Alnabelsi et al., 2015;Power & St-Jacques, 2014;Tseng et al., 2019). Research exploring the opportunities for the educational sector in the United Arab Emirates (UAE) during the COVID-19 pandemic was undertaken by Abdallah and Alriyami (2022).A survey of 100 teachers from 20 higher education institutions in the UAE was conducted to determine the effectiveness of online learning tools and their ability to replace face-to-face interactions with students.The research method included primary and secondary data analysis, and they administered a questionnaire comprising nine questions to teachers.The findings suggest that only a small percentage of teachers agreed with the inclusiveness and ease of remote learning technologies use, and a significant number disagreed that the technologies used stimulate creativity and have an evidence base.The study highlights the need for significant reforms in distance education for its adaptation in the post-COVID-19 period.We have observed similar consequences of school shutdowns in education systems worldwide. Research Model The study tests the UTAUT model for determining the influence of technology (webinars) usage on performance expectancy, social influence, and effort expectancy on the intention of the student to use technology (webinars).Not only this, but in the original UTAUT model, the moderator variables were found to be age, gender, experience, and voluntary use (Venkatesh et al., 2003).However, researchers have scarcely explored the individual effect of the factors on the UTAUT.Venkatesh et al. (2016) has emphasized this, which showed that surpassing the moderating effects can cause an adverse impact on the UTAUT generalizability.The study has adopted the guidelines of UTAUT contextualization, as showed by Hong et al. (2013) for a blended learning environment.The researcher also suggested incorporating context-based factors for better results.Therefore, the present study has considered two factors, i.e., outcome and satisfaction, to bridge the gap concerning the association between intention and behavior, as recommended by Tate et al. (2015). Empirical Investigations Using technology, as indicated through the modern standards of educational institutions, has diverted the interest of various educational institutions, while eliminating the use of books in the classroom.Webinar is one of the advanced technological tools which provides an opportunity to carry out physical classes to an online forum by using a videoconferencing system.This system or term is new to the field of education.The main idea behind webinars is to conduct online seminars.But it is not just the only event to carry out, through webinars meetings, demonstrations, conferences, training, or teaching and many other similar events can be carried out interactively (Gupta & Sengupta, 2021).Young et al. (2002), suggested a "3I Framework theory" which states that we should give students an opportunity to interact with videos.Using audio-visual technology has long been there in the education sector but with the improvement of information and communication technology, its part has become more active and massive today (Gupta & Sengupta, 2021).Rashid and Asghar (2016) illustrated that easy access to computers for different students helps in developing a positive attitude regarding the use of technology and provides competitive results.Drossel et al. (2017) emphasize that teachers having a negative attitude toward the use of technology are unlikely to integrate the use of technology in classrooms.Afari and Khine (2016) in their study, explain the use of mobile learning technology in Taiwan and selected 320 respondents to investigate the issue.They found variables related to the behavioral intention other than social influence are highly important.Besides this, individuals with high-level performance expectancy were prone to use mobile technology to provide efficient results. Almekhlafi (2016) illustrated that UAE stands among those countries that are integrated into high-level efforts in implementing ICT in the modern education system.In various universities of Abu Dhabi, webinars have been integrated as an effective strategy to enhance the teaching and learning process.Besides this, various faculty members use webinars to conduct online classes, assessing students' performances, conducting assignments to increase the students' interaction towards highly beneficial communication tools in the form of modern technology.Jabeen et al. (2018) investigated the use of webinars in the modern educational system of Abu Dhabi through the Technology Acceptance Model (TAM).It further illustrated different factors to predict students' intentions and attitudes regarding its use.The findings of the study illustrated that perceived technological literacy and e-satisfaction was found as an important determinant in implementing the use of webinars in educational institutions.Heflin et al. (2017) suggested that educational strategies that are based on the use of modern technology increase the ability of critical thinking among students.This further creates a positive impact on the decision-making process and problem-solving skills.According to McKnight et al. (2016), the use of emerging technologies provides open opportunities for students and teachers to enhance the learning process by incorporating visual aids.This has contributed to the developments of useful websites and software that support multiple teaching and learning strategies in classrooms.Goundar (2014) conducted a study illustrating the modern technology provided by ICT is penetrating and thus demanding a high level of individual attention.The ecology of the learning environment has not changed from learning the material to the source of distraction.ICT has changed previously incorporated educational standards, incorporating learning and writing skills.Findings of the study illustrated that students experienced high-level distractions because of the constant use of ICT devices in classrooms for non-academic purposes.Kryukov and Gorin (2016) demonstrated that teacher serves as the principal source of knowledge for students, students with the frequent use of technology become less dependent on their teachers.The study further illustrated that implementing modern technology alters learning for various teachers.Use of digital technology in modern times may not deliver learning competitive to face to face learning.Salloum and Shaalan (2018) further investigated the use of technology in various universities through the TAM model.The model serves as an important determinant in influencing the decision-making process to the use of e-learning.This further provides an emphasis on the use of visual and audio aids, animations, and experiment-based videos to stimulate greater interest and learning feasibility among individuals. Overall, the literature review provides a comprehensive understanding of the UTAUT model and its relevance in investigating the use of technology in the education sector.The empirical analysis highlights the potential benefits and drawbacks of technology, emphasizing the need for a balanced approach in integrating technology in the learning process. Hypotheses Based on the discussions, we developed the following hypotheses: H1: There is a significant relationship between performance expectancy, effort expectancy, and social influence on the behavioral intention of the student.H2: There is a significant relationship between facilitating conditions and voluntariness of use of the behavioral intention of the student.H3: There is a significant relationship between age and gender on the behavioral intention of the student. METHoDoLoGy Study Design The study used a cross-sectional research design to achieve the determined objectives.It used a quantitative approach for analyzing the students' perspectives regarding the implementation of the use of modern technology in classrooms.The rationale for using this research design is based on its effectiveness for concluding effective and quantifiable results, as established in previous studies. Study Population and Sample The population of the current study involves students that are using technology in classrooms.Participants of the study involved 103 students and 37 faculty members at a university in Abu Dhabi, to collect data regarding their perceptions of the use of technology in the classroom.However, samples were collected through a purposive sampling technique based on the determined inclusion criteria (Table 1). We demonstrate the demographic details of the respondentsare presented in Table 2. Based on the responses, we found that the majority of the sample comprise female population, i.e., 62, whereas the number of males was 41 (1.6019 ± .49189).The age analysis of the respondents depicts that most of them are in 17 to 22 years age bracket (44), following the respondents who were above 30 years of age (39) while only a few were aged between 28 to 32 years (2.3883 ± 1.36650).Also, the majority of the participants belonged to first year (30), followed by students in the fourth year (26), and in the second year (22), while only 11 students from the third year took part in the study.Also, the sample mostly comprises respondents who were students (67) in contrast to the number of faculty members (37) (Mean, 1.6442, Std.Deviation, 0.48106). The discipline evaluation of the participants indicated that most of the participants were associated with Arts and Science College (50), following students in Engineering College (29) while only a few Inclusion Criteria Exclusion Criteria Students enrolled in classes using webinar Students not enrolled in classes using a webinar Provided Consent Did not provide consent from the business college ( 14).Most of the responses revealed participants were involved in daily usage of the internet (93), while 3 participants used it every week while 1 participant used a few times in a year.Close responses were found for yes and maybe when considering the intention to use the online course, i.e., 45 and 44, respectively, whereas only 12 responded no. Data Collection Data is collected through a survey based on a questionnaire that is further divided into two parts.The first part of the questionnaire includes demographic information of the participants, while the second part of the questionnaire collected information related to technological acceptance.Overall, the questionnaire consists of 58 items, which required about 20 minutes to be completed.Except for the demographic variables, all the other items used a 7-point Likert scale where one represents strongly disagree while seven represents strongly agree.The Venkatesh et al. (2003) questionnaire of 36-items was adopted while the Fillion (2005) study guided gathering responses concerning outcomes such as performance and satisfaction. Ethical Considerations The study gained ethical approval from the institutional review board at a university in Abu Dhabi. Further, the study attained a written consent from the participants while also ensuring the confidentiality and anonymity of the collected data.We will keep information collected in this study securely. Data Analysis We analyzed the data using IBM SPSS version 20.0.We employed descriptive statistics to analyse the categorical variables and Structural Equation Modelling (SEM) to assess the continuous variables. RESULTS Table 3 presents the convergent validity analysis with the use of average variance extracted (AVE) measures.Based on Chin (1998), the AVEs values over 0.5 emphasized the correlation among the variables.We observed that values of all the factors lie above 0.5, such as for performance expectancy it is (0.700) while for effort expectancy it is (0.685), social influence (0.585), and for facilitating conditions it is (0.734).The findings in Table 3 provide evidence of the discriminatory validity of the scales.Table 4 provides the results as per the Fornell and Larcker (1981) criterion, where the diagonal elements present the AVE square root and the correlation among the latent variables.The values of the variable represent a higher value which established the attainment of the determined Fornell and Larcker (1981) Criteria.The Heterotrait-monotrait (HTMT) ratio measures the discriminatory validity of the variables (Henseler et al., 2015).It shows the sensitivity ratio and the specificity other than the one presented by Fornell-Larcker and cross-loadings.The HTMT ratio must be lower than 0.90 for establishing the discriminate validity (Table 5). Path analysis presents the significant relationship that prevails between the study variables (Table 6).It shows that there is a strong relationship among the study variables (p-value <0.05), except for voluntariness of use and facilitating conditions with behavioural intentions (p-value 0.61 and 0.10, respectively).It shows that performance expectancy, effort expectancy, and social influence with behavioural intentions (p-value < 0.05).Similar is the impact of age and gender on the behavioural intention of the student, as depicted through the achieved p-value, i.e., 0.05 and 0.04.Therefore, we accepted H1 and H3 and rejected H2. DISCUSSIoN The study provided a detailed analysis of the use of webinars in classrooms and the perceptions and intentions of using them in Abu Dhabi universities.The study considering the UTAUT model and the use of technology (webinars).We primarily composed the sample of the study of students in contrast to faculty members.Most of the students belonged to the first year of their academic program, which was followed by second-year students.However, a minimal number of students extent took part in the second and third years.Besides this, most of the participants belonged to Arts and Science, while others belonged to Engineering and Business colleges, respectively.We found the demographic variables such as age and experience to affect the intention to use among the students.Celik (2016) found that usage intention is moderated by the age and experience of the individual, which corroborated our study findings. The results showed that the performance expectancy helps academicians to encourage the use of webinars in higher education.This is consistent with the findings of Khechine et al. (2013) which suggest that webinars can induce students' willingness to use technology; however, it should be used with care.Students' intention of using the internet is directly associated with their perceptions of using technology.Also, most of the students reflected a positive attitude regarding their perceptions of using online courses.Badri et al. (2016) agreed with the idea that technology serves as the major driver in supporting the intention of using online courses.They further suggested that perceived usefulness and perceived ease of use create a direct impact on students' behavioral intention of using technology.Another considerable part of the study includes users' intention regarding the skillfulness of technology, which has shown a significant positive association between them.Similarly, students' intention of using technology increases with their perceptions about the usefulness of the technology.The findings in Scherer et al. (2015) supported the idea of the present study, where they demonsrated that a high level of perceived usefulness is associated with levels of technological skills and acceptance among users.Skillful students using technology usually have high levels of self-efficacy regarding the use of technology.This shows a positive relationship between perceived usefulness and individual perceptions regarding the use of technology and its acceptance.Khechine et al. (2016) supported the findings stating that they link the effort expectancy to the behavioral intention.Most of the students bring their technological devices to support their learning and understanding related to everyday classroom knowledge.Students' satisfaction related to the use of technology was associated only with the competency of technology in fulfilling their needs.This promoted students' collaborations and peer learning largely in practice.Lecturers, however, supported similar perceptions where technology is highly important for integrating efficient educational practices.This pinpoints the fact that both students and teachers are largely to support using technology (webinars) in classrooms.Percival and Claydon (2015) provided similar results according to which student's intention of using technology (webinars) was dominant during classroom practices. The study further illustrated that the use of technology among students was at an average of four to five hours per day in different classroom activities.Among them, some of the major tasks include managing classroom learning and coursework completion.Jabeen et al. (2018) indicated that technological skillfulness and behavioral intention are highly significant factors for students regarding the use of technology in the classroom.Certain platforms associated with the use of technology, such as blackboards, are highly convenient for students because of several reasons.First, easy access to course materials requires minimal efforts.Second, there is enhanced feasibility of discussion of general classroom-related issues regarding assignments, projects, and lectures.Use of technology reduces the time for grading purposes, providing an opportunity for students for grading purposes.This is only possible because of students' competency in acquiring basic technological skills, that it is a major factor in promoting the use of technology in classrooms. The findings of the study emphasize that teachers must not exert pressure on the students to adopt the webinar technology.Instead, the teachers must promote the idea of the voluntary use of technology, which may increase their intention to use.Exerting pressures is likely to make students rigid and resistance.The academicians must emphasize the technology usefulness and provide the students freedom for using webinars.The teachers must market webinar use and its potential benefits, aligning their work with generation Z preferences, such as through different social medium channels.The instructors must use the webinar system rather than using the classroom blackboard.We encourage that classroom educational intervention strategies must use pedagogical activities, more interaction, and more breaks.Similarly, teachers must also advise students on how to optimize their learning to use technology.The study further recommends future research to use the UTAUT model as proposed by Venkatesh et al. (2012).This helps in expanding the research area further given its integration of price value, hedonic motivation and habit.This understanding helps in promoting more secure strategies for promoting students to adopt technology and improve their learning outcomes. Overall, the study contributes to the existing body of literature on the use of technology in education and provides valuable insights for academic institutions and policymakers to promote the adoption of technology for improving the learning outcomes of students. Study Limitations The study only considered universities of one emirate of the United Arab Emirates due to time constraints, which is therefore the limitation of the study. Future Recommendations The future researchers can consider other emirates to compare students' opinion related to different characteristics of webinars.Future researchers can use data envelopment analysis to find the efficiency of webinars in different emirates of UAE and identify the efficiency determinants through Tobi regression or the ordinary least method.Different statistical tools, such as factor analysis, multinomial logistic regression, principal component analysis and composite index, may be useful for the critical analysis of students' perceptions. CoNCLUSIoN AND LEARNED LESSoNS Use of technology (webinar) serves as a valuable idea in providing advanced education, specifically at a university in Abu Dhabi.This provides significant importance towards the usage of technology (webinar) among students and considers factors that are significant to users.The main idea of this study involves the usefulness and intentions of students for integrating technological based learning (webinar) at a university in Abu Dhabi.This includes various factors depending on different dimensions of learning to achieve targeted goals.Our study results are crucial, as they show students' perceptions and intentions regarding the use of technology in the classroom. The study is significant as it examined students' intentions, attitudes, and perceptions regarding the use of technology (webinar) at a university in Abu Dhabi.This provides an opportunity for future researchers to explore the performance of the university after successful integration of technology in classrooms.The study provides an opportunity to examine students' level of satisfaction and dissatisfaction when learning is incorporated using technology.This may help educational experts to design the most appropriate method of teaching and learning to use technology.The results can help policymakers in integrating the use of webinars in educational institutions while considering the factors.The analysis of the given factors may provide useful insights regarding the efficient usage of webinars in education.Thus, the study is useful in contributing to the development of information and communication technology system in education.This will help educational experts to build selfefficacy among learners regarding the use of technology. There are five principal lessons: • Technology, particularly webinars, can provide advanced education and is a valuable tool in the classroom.It is important for educational institutions to integrate technology into their teaching methods to provide better learning experiences for students.• Educational experts should consider students' perceptions and intentions regarding the use of technology (webinar) in the classroom when designing teaching and learning methods.This can help educational experts to provide the most appropriate method of teaching and learning that suits students' needs. • Integrating technology in the classroom can provide an opportunity for future researchers to explore the performance of the University and the level of satisfaction and dissatisfaction among students.This can help policymakers to design effective policies for integrating the use of webinars in educational institutions. Figure 1 depicts the pictorial representation of the UTAUT model as deployed in the present study.The independent variables in the study include (Performance expectancy, effort expectancy, social influence, and facilitating conditions), while the dependent variables are BI and UB.Age, experience, gender, and voluntariness to use are the moderating variables. Figure Figure 1.Research model Table 6 . Path analysis • The study contributes to the development of information and communication technology systems in education.This will help educational experts to build self-efficacy among learners regarding the use of technology.•Educational institutions should consider various factors that are significant to users when integrating technology into the classroom.These factors can depend on different dimensions of learning to achieve targeted goals.She firmly believes that good is the enemy of great, and she works tirelessly to inspire and motivate others to aim for greatness in everything they do.She is a highly accomplished and respected education professional who has made significant contributions to the field of education.Her dedication to excellence, her passion for teaching, and her commitment to research and innovation have earned her a welldeserved reputation as a leader and role model in the education community.As an active researcher, Dr. Asma has published several articles on topics related to her research interests, which include educational leadership, school improvement, and education quality.She brings her rich practical experience and knowledge to bear in her research and is widely respected for her insights and contributions to the field.Osama Ismail is a dedicated e-mobility engineer turned educator who is passionate about teaching engineering.Osama graduated from the University of Antwerp with a master's degree in Sustainable Automotive Engineering and then went on to work for various automotive companies.After many years in the automotive industry, Osama took the initiative for education.He wanted to share his love of engineering and technology with the next generation, so he decided to become an automotive engineering teacher alongside his research in the e-mobility sector.Rasha Khalil holds a Master's Degree in English Language and Literature.Rasha's teaching philosophy is to improve teaching and learning by providing empowered, highly qualified teachers and leaders.She believes that in education good is the enemy of great.Rasha teaches English for high school in a school of an American curriculum.She is the Head of the Girls' Section where the teaching and learning process should be ensured to the maximum in a safe, positive and motivating environment.Ahmed Alkaabi is an assistant professor in the Foundations of Education Department--College of Education at United Arab Emirates University.He is currently serving as the Director of the Emirates Institute for Learning Outcomes Assessment at UAEU as well as Coordinator of the Master of the Educational Innovation Program.His educational qualifications include a Ph.D. in Educational Administration and Policy with an emphasis on supervision from the University of Georgia-Athens, USA (2019).Dr. Alkaabi also earned two master's degrees--the first in School Leadership from United Arab Emirates University in 2014 and the second in Educational Administration from Ohio State University, USA in 2015.During his decorated journey, he was the recipient of two distinguished academic awards: the Ray Bruce Award in 2017 for his dedicated work and projects in the field of instructional supervision, and the Faculty Award in 2019 for his accomplishments in the Educational Administration and Policy Doctoral Program at the University of Georgia.Dr. Alkaabi has contributed several scholarly articles in the field of educational leadership, which have been published in internationally recognized journals indexed by Scopus.His research interests reflect his expertise in school leadership, specifically in the areas of supervision, evaluation, Asma Abdallah is an assistant professor in the Foundations of Education Department--College of Education at United Arab Emirates University.She is an experienced and highly qualified education professional with a passion for teaching and a dedication to promoting excellence in the field of education.She holds both a Ph.D. and a Master's degree in Education Administration and Leadership, and has worked as a teacher, vice principal, principal, and school's education improvement consultant throughout her career.In addition, she is a certified school inspector, which reflects her deep understanding of the importance of educational quality and standards.Dr. Asma is known for her dynamic and engaging teaching style, and she is committed to helping students and educators alike to achieve their full potential.induction, professional development, and data-driven decision-making.	2023-05-14T15:16:17.155Z	2023-05-12T00:00:00.000	{ "year": 2023, "sha1": "92dd0e156249ed85e7b5416059377584c2d7b9ff", "oa_license": null, "oa_url": "https://www.igi-global.com/ViewTitle.aspx?TitleId=322793&isxn=9781668478998", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "6eacc2bccdc075be864c2249a4618142b4168ed4", "s2fieldsofstudy": [ "Education" ], "extfieldsofstudy": [ "Computer Science" ] }
14638546	pes2o/s2orc	v3-fos-license	The selection of experts evaluating health projects for the EU Sixth Framework Program Aim The Framework Programmes for Research and Technological Development (FP) are the European Union’s funding programmes for research in Europe. The study analyses the features of external experts involved in evaluating the research proposals in FP6 (years 2003–2006) in the area of Life Sciences. Subjects and methods Experts were analysed with respect to nationality, gender, organisational affiliation and rotation. The correlations between the number of experts by nationality and scientific research indicators were also explored. Result Experts from 70 countries participated, with 70% coming from 10 countries. The gender composition was relatively stable, with approximately 30% of female experts. The majority of experts came from higher education establishments (51%) and 12% from industry. About 40% of experts participated in the evaluation process two or more times. The number of experts by nationality was linearly correlated with gross national income (r = 0.95, p < 0.0001), population (r = 0.91, p < 0.0001), and number of research publications in health sciences (r = 0.93, p < 0.0001). However, using multiple linear regression analysis, only gross national income had partial regression coefficients significantly different from zero (p = 0.017). The observed value of experts for Italy (312) and Belgium (155) were higher than predicted by this regression model (231 and 71 respectively). Conclusion The expert panels involved were balanced with respect to nationalities, whereas the gender distribution was lower than the target. There was a satisfactory degree of rotation of experts between evaluation rounds. The percentage of experts from industry was lower than expected. Introduction The Framework Programmes for Research and Technological Development (FP) are European Union (EU) funding programmes designed to address major basic and applied research projects in Europe. The budgets of the FPs have increased steadily since the first FP was introduced in 1984, and the current seventh FP (FP7, 2007(FP7, -2013 has an average annual budget of more than 7 billion € (Andrè 2006;European Commission 2007). The specific objectives and funding instruments have changed between the consecutive FPs, but continue to keep a strong emphasis on supporting collaborative, multidisciplinary research between research teams in different countries. The FPs are open to all types of research organisations, promoting scientific collaboration between researchers in universities, industry, small and medium enterprises (SMEs), research centres and public authorities (European Commission 2007). Research grants from the FPs are awarded on the basis of applications submitted to specific calls for proposals. The eligible applications undergo a rigorous peer review process in order to select the best proposals for potential funding by the European Commission (EC). The final funding decisions are made using a range of criteria, including excellent scientific quality, potential scientific and societal impact, quality of the management team, portfolio balance, and relevance to the overall objectives of the FP. As a consequence, the most important factor for the funding decision is the result of the peer review by independent experts (European Commission 2008a). In the thematic area of health, proposals undergo a first review in a remote step, where a number of experts are individually evaluating and scoring the proposals. In a second phase, the experts meet in consensus groups to reach agreement regarding the scoring and ranking of submitted research proposals. This results in a prioritised ranking list of the submitted proposals, and a recommendation to the EC for funding. The peer review of proposals by external experts is a crucial step in the funding decision process of the EC. It is therefore critical that the evaluation process is carried out according to the highest possible standards with respect to scientific competence, fairness, and integrity. The roles and actions of independent experts are put forward by EU guidelines (European Commission 2008a, 2005, 2006a, 2002a. The EC maintains the database of potential experts who can be called upon to assist with peer reviews. The database is open to auto-registration, with no formal assessment or validation (CORDIS 2007). The EC encourages research institutions to submit lists of proposed experts who could be called to assist the Commission in the proposal evaluation process (European Commission 2006a), and world-class experts to register as experts. In any case, experts participate in their own personal capacity, rather than representing the organisation with which they are affiliated (European Commission 2006a). The information required for registration of candidate experts is structured in order to have an easily searchable database, where information is selected predominantly from pre-defined lists ("closed" questions). Every time a round of reviews is scheduled, experts are drawn from the database using a variety of search criteria such as keywords related to the research area concerned. Experts are then analysed individually, and the EC draws up a list of appropriate independent experts, using the following selection criteria: (1) a lack of conflict of interest; (2) an appropriate range of competencies, expertise and experience; (3) a reasonable balance between academic and industrial expertise; (4) a distribution of geographical origins; and (5) a reasonable gender balance (European Commission 2008a;European Commission 2006a;European Commission 2002a). Furthermore, a principle of "regular rotation" is applied to the selection of experts in order to avoid that the same individuals bring the same views, expertise and arguments within evaluations. The whole process is monitored externally and reviewed for all calls for proposals by independent observers (Catenhusen, Grimaud and Horvat 2007;Cahill and Horvat 2008). The aim of this paper is to review and assess the pool of external evaluators with respect to their country of origin, type of affiliation, gender and renewal of experts called to participate to panels (rotation). In addition, we examined as to whether the geographical distribution of experts could be correlated with national indicators such as national size, wealth and breadth of scientific research. The analysis has been limited to the activity area 1 "Life sciences, genomics and biotechnology for health" for FP6 (European Commission 2002b). Preparation of experts' data Lists of experts that participated in the evaluation of proposals were downloaded from the CORDIS website, where they are posted yearly (CORDIS 2006). All lists for the activity area 1 "Life sciences, genomics and biotechnology for health" of FP6 (years 2003-2006) were downloaded and analysed. Evaluations for FP6 calls from year 2002 were held in 2003. For each expert, the following basic information was available in the published lists: name, gender, nationality, organisation and organisation type. Data were reviewed to ensure consistency from one year to the next, and for possible errors, including screening of the names and affiliations of experts for variants of the same name. An extensive effort was also made to review and validate the type of organisation which the experts were affiliated to. All organisations where the word "university" appeared in the organisation name were, for example, classified as "higher education establishment". National research indicators and statistical analyses The possible correlation between the number of experts by nationality, scientific research indicators or other features of nations was explored. The indicators included: (1) population; (2) gross national income (GNI) (World Bank 2010; King 2004); and (3) numbers of published articles (years 2000-2004) in fields pertaining to health sciences-biology, biochemistry, clinical medicine, immunology, microbiology, molecular biology, genetics, neurosciences and behaviour, pharmacology, psychology and psychiatry (Thomson Reuters 2010). Descriptive statistics were presented as medians with interquartile ranges (IQR). Where indicated, statistical significance was tested with chi-square test on frequency tables (Analyse-it for Microsoft Excel, Version 2.11, Analyse-it Software Ltd., 2008). Pearson's correlation coefficient r and bivariate or multiple linear regression were computed using SPSS Statistics 17.0 (SPSS Inc., USA). Nationalities and gender of experts During the 4-year period of FP6, the EC used a total of 3,057 expert evaluations to assess research proposals submitted for funding in the area of Life Sciences. In 2003, 799 expert evaluations were used, 983 in 2004, 672 in 2005 and 603 in 2006. Experts came from 70 different nations but 70% of all experts came from just 10 countries, and 22 nations contributed comprising 90% of the experts (Table 1). During the course of FP6, 12 new countries (EU-12 new) joined the EU. We therefore examined whether the proportion of experts from the new Member States increased during the period, but no clear evidence of this was found (Fig. 1 The overall percentage of females among experts was 27%, and the gender distribution was stable throughout FP6. Participation of female evaluators was 24, 28, 29 and 28% respectively, for each year from 2003 to 2006. The gender proportion was more varied when analysing data by groups of countries. The percentage of females among experts from the EU-12 new countries was significantly higher than among experts from EU-15 (39% vs. 27%, p< 0.0001). For nations with at least 10 expert evaluators (the 34 nationalities listed in Table 1), the percentages of female experts ranged between 5 and 91%, with a median value 31% (IQR=22%). Thirty-one per cent is also the value for all 70 nationalities (IQR=40%). Table 2 summarises the affiliations of the recruited experts. More than half of all experts (50.6%) came from higher education establishments, while a quarter (25.5%) came from a variety of public research institutions. Experts from the private for-profit sector, including large companies, SMEs, consultancy firms and service providers, represent 11.9% of all evaluators. Table 3 shows the nationality of experts coming from different types of organisations in the EU countries (United Nations Statistics Division 2010). The highest percentage of university affiliations was found among the experts from northern EU (57%), whereas this group had the lowest percentage of experts from other areas of the public research sector (19%). Southern EU provided the highest percentage of experts from public (29%) and private non-profit organisations (12%). The highest percentages of experts from private for-profit organisations were among experts from northern (13%) and western EU (15%), whereas 9% of experts from southern and eastern EU came from this sector. Rotation The total number of expert evaluations used in FP6 was 3,057. However, the number of different experts was smaller as many individuals took part in more than one round of evaluations. When data was corrected for repeated participation, the number of different experts involved in FP6 for the area Life Sciences was 1,982. About 40% of experts were involved two or more times, while a large majority (1,192 experts, 60.1%) participating only once; 540 experts (27%) twice, 215 experts (11%) three times, and 35 experts (2%) appeared in all 4 years. Correlation with national indicators We performed a bivariate correlation analysis for a set of research and wealth indicators, and found that the number of experts by nationality was linearly correlated with gross national income (GNI: r=0.95, p<0.0001), population (r= 0.91, p<0.0001), and number of research publications in health sciences (r=0.93, p<0.0001) in the EU countries. However, using multiple linear regression to investigate the independent effects of individual factors, only GNI had partial regression coefficients significantly different from zero ( Table 4). Analysis of standardised residuals, applied after multiple regression, showed that the observed value of experts for Italy (312) and Belgium (155) Fig. 2 for illustrative purposes. Discussion Peer-review evaluations are a core element of funding decisions by most major research funding agencies. The overall approach of the evaluation process for the FP has been favourably reviewed by independent observers on recurring occasions (Catenhusen, Grimaud and Horvat 2007;Cahill and Horvat 2008). However, little statistical information about the pool of experts involved in the evaluations has been available until now. There are very few quantitative comparisons in the evaluation process between different agencies, national and international. The quantitative information that is available is generally part of larger studies in which the evaluation process is not the core focus. Comparison of different evaluation systems is difficult, if not impossible, because they are established according to varying approaches and objectives. The present study sought to provide quantitative information about the evaluation system and, in particular, the characteristics of the evaluators involved in the EC peer-review process. It is a stated goal of the EC to increase the number of women participating as experts for the FP evaluations towards a suggested target of at least 40% (European Commission 2008b). The hypothesis was that this would potentially lead to an increased number of female participants in FP7 proposals. The present analysis found that the overall percentage of female evaluators was 27%. This reflects an evident gender imbalance as a whole in science and research where women are clearly under-employed (Klinge 2008). Across the EU, 29% of researchers are women, although the proportion of female researchers in medical sciences in the higher education sector of the EU-25 is 40% (European Commission 2006a). In 2006, the percentage of female evaluators of all evaluation panels of other themes of FP6 was 34% (European Commission 2008b). The gender disproportion which was shown in the present analysis had been repeatedly noted in other evaluations of proposals (Neugebauer 2006;Wenneras and Wold 1997). As a comparison, the overall percentage of women among the core reviewers in the US National Institute of Health (NIH) in the years 2003-2006 (corresponding to FP6) was 27.4% (Dumais, Lindquist and Malik 2004). The shortage of women in the EC's expert database forms a major obstacle for ensuring the gender balance of expert evaluation panels, although this situation has been significantly improved in recent years. In 2000, women accounted for only 16% of the experts registered in the EC's expert database, whereas 26% of the experts in the database were women in 2006 (Laurila and Young 2001). It has been noted that in some cases when an expert's evaluation experience is necessary, a preference was given to previously selected candidates; this may have formed a barrier against the inclusion of women as newcomers (Laurila and Young 2001). Furthermore, while gender balance is one of the criteria for evaluator selection, it is applied after the criteria of knowledge and geographical origin (Information 2001). An important feature of the evaluation process is the type of organizations to which the expert evaluators are affiliated. Experts are recruited as individuals, but their affiliation may, nevertheless, indicate the type of experience and background they bring into the evaluation. Experts from higher education establishments comprised more than half of all experts, while an additional 25% came from various types of public research institutions. As a comparison, the percentage of reviewers from universities in the NIH panels (2003-2006, standing and non-standing reviewers) was 81.7% (Dumais et al. 2004). The overall percentage of experts coming from the private for-profit sector ("industry") was 12% in FP6, but interestingly it increased from 10% in 2003 to 17% in 2006. This should be seen in the light of the stated target of the EC to involve industry, and particularly SMEs, in research activities funded under FP6. The increased number of experts with industry background could be a consequence of more research proposals with industry involvement, and thus requiring industry expertise for the evaluation. Independent observers stated that the contribution of industrial experts to the evaluation process was positive, increasing the diversity and perspectives of the evaluators especially with regard to the potential impact of the proposals (Cahill and Horvat 2008). Use of the same experts in two or several rounds of proposal evaluations is considered a strength as well as a weakness. On one hand, it secures a certain degree of stability and consistency between separate evaluations; while on the other hand, the same scientific judgments may prevail in several evaluations to the detriment of new approaches. Funding agencies approach this issue differently. The NIH review system includes a core group of reviewers (roughly 20% of the total) and non-standing temporary reviewers who provide additional expertise needed to enhance the review of applications received in any given review round. There is no rotation schedule for non-standing reviewers, who are contacted to serve on an "as needed" basis (Dumais et al. 2004). In the EC system, a principle of "regular rotation" is applied, and a high turnover rate was consequently observed over the years in FP6. However, given the wide range of topics, it may not always be possible to find new evaluators with sufficient and specific expertise in a niche area. Experts with specific knowledge in a niche area, as well as broad expertise of their field, are therefore sometimes invited several times. However, only 2% of experts, corresponding to a total of 35 individuals, participated in evaluations in all 4 years of FP6. A more detailed analysis revealed that these experts were well-recognized experts in their field, having an average publication track of more than 40 publications in peerreviewed journals during the last 10 years. Many of them had a highly specialized knowledge in a niche such as intellectual property right or vaccine manufacturing, whereas others displayed an uncommon combination of personal and professional background such as female scientists in industry or senior scientists from the new EU Member States. The number of experts called from the different EU nations appeared to be correlated to country size (population), wealth (GNI) and scientific production (research articles). Multivariate analysis showed only a statistically significant correlation with wealth of the EU nations, meaning that the number of experts from a given EU country could be predicted by the GNI of the country. Only two countries, Belgium and Italy, contributed with a slightly larger number of experts than predicted by the GNI, whereas all other EU countries fell within the 95% confidence level. Geographical proximity and budget considerations could explain the somewhat larger representation of experts from Belgium, whereas a larger share of experts from Italy could be due to do a larger share of Italian candidacies in the expert database. Table 1 are labelled Conclusion In conclusion, we found that over two thirds of the experts used for the evaluation of research proposals in the thematic area of "Life sciences, genomics and biotechnology for health" of FP6 came from 10 countries, and the number of experts from a given EU country seems proportional to the GNI of the country, except for Belgium and Italy. The majority of the experts came from higher education establishments, whereas 12% came from industry. The group of experts had a relatively constant and low percentage of women (27%), with the percentage comparable to other large research funders. The role and level of expert rotation was difficult to analyse due to the limited study period, but only a small percentage of experts were called on multiple occasions. It should be kept in mind that the present study was limited to a single thematic area, and the results can, therefore, not be considered representative for the EC research evaluation system for FP6 as a whole; however, some indications might be of use in future selections, especially in terms of gender proportion and the contribution of academic versus industrial expertise. Evaluations can and should be studied quantitatively. This approach can be an important support tool to the extensive qualitative assessment and external auditing of evaluations already used by the EC, and can help highlight merits and weaknesses. Further study on quantitative aspects of the research evaluation process such as more accurate comparisons between different evaluation systems carried out by various research agencies, is necessary.	2014-10-01T00:00:00.000Z	2011-02-08T00:00:00.000	{ "year": 2011, "sha1": "0964b27bf22cb3f6bc93866ea18772d9397f10f8", "oa_license": "CCBYNC", "oa_url": "https://link.springer.com/content/pdf/10.1007/s10389-011-0395-5.pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "0964b27bf22cb3f6bc93866ea18772d9397f10f8", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Political Science", "Medicine" ] }
211546593	pes2o/s2orc	v3-fos-license	Stress field orientation controls on fault leakage at a natural CO 2 reservoir . 10 Travertine deposits present above the St. Johns Dome natural CO 2 reservoir in Arizona, USA, document a long (>400 ka) history of surface leakage of CO 2 from a subsurface reservoir. These deposits are concentrated along surface traces of faults implying that there has been a structural control on the migration pathway of CO 2 rich fluids. Here, we combine slip tendency and fracture stability to analyse the geomechanical stability of the reservoir-bounding Coyote Wash Fault for three different stress fields and two interpreted fault rock types to predict areas with high leakage risks. We find that these areas coincide with 15 the travertine deposits on the surface indicating that high permeability pathways as a result of critically stressed fracture networks exist in both a fault damage zone and around a fault tip. We conclude that these structural features control leakage. Importantly, we find that even without in-situ stress field data, the known leakage points can be predicted using geomechanical analyses, despite the unconstrained tectonic setting. Whilst acquiring high quality stress field data for secure subsurface CO 2 or energy storage remains critical, we shown that a first order assessment of leakage risks during site selection can be made 20 with limited stress field knowledge. Introduction The successful subsurface storage of fluids in sedimentary basins is key for GeoEnergy technologies such as Carbon Capture and Storage (CCS), cited as a cost-effective tool for climate change mitigation, or for energy storage, required to balance the intermittency of future energy systems relying on renewable sources (Alcalde et al., 2018;Matos et al., 2019;Scott et al., 25 2013). The integrity of such engineered subsurface storage sites is controlled by a range of geological, geochemical, and geotechnical factors. One major concern is that impermeable caprock seals may be bypassed by faults and naturally occurring, or induced, fracture networks which can form preferential fluid pathways. These could provide conduits for fluid migration, potentially leading to the rapid migration of the stored fluid (e.g., CO2, H2, methane) to shallow aquifers or the atmosphere (IPCC, 2005;Shipton et al., 2004;Song and Zhang, 2012). Indeed, selection criteria for subsurface storage sites commonly 30 cite the need for minimal faulting and/or low permeability faults intersecting or bounding the storage site (Chadwick et al., 2008;IEA GHG, 2009;Miocic et al., 2016). However, within sedimentary basins, which are key targets for geological storage of fluids, faults will occur naturally close to or within a storage complex and thus predictability of whether a fault will act as barrier to fluid flow or not is key for an accurate risk assessment. Whether a fault zone is sealing or non-sealing is dependent on the structure and composition of the fault zone and the mechanics 35 of the faulting (Faulkner et al., 2010). In a widely used simple conceptual model for fault zones in siliciclastic rocks, strain is localized in the fault core that is surrounded by a damage zone of secondary structural discontinuities. Fault zones can have a single high-strain core (Chester and Logan, 1986) or contain several cores (Choi et al., 2016;Faulkner et al., 2003). The damage zone and the fault core have contrasting mechanical and hydraulic properties, with the fault core often being rich in phyllosilicates which typically have low permeability. Contrastingly, open fractures in the damage zone can have a 40 substantially higher permeability than the host rock, if not diagenetically cemented (Caine et al., 1996;Cappa, 2009;Faulkner and Rutter, 2001;Guglielmi et al., 2008). Lateral fluid migration across the fault zone is thus controlled (1) by the permeability and continuity of the fault gouge/rock within the fault core(s), which is dependent on the host rock composition, shear strain and faulting mechanism, as well as (2) the juxtaposition of strata across the fault (Yielding et al., 1997). Inversely, vertical fluid migration is governed by fracture permeability in the damage zone (Davatzes and Aydin, 2005). 45 A significant amount of research has focused on understanding the mechanisms and parameters that control the composition, and continuity of fault gouges as well as their permeability for different fluids as they have the potential to form effective seals (Karolytė et al., 2020;Lehner and Pilaar, 1997;Lindsay et al., 1993;Miocic et al., 2019b;Vrolijk et al., 2016). The damage zone permeability is controlled by the permeability of the host rock, the presence and geometric composition of macro-scale fracture networks and deformation band networks which decrease in frequency with increasing distance from the fault core, 50 as well as burial history, cementation and in situ stresses (Mitchell and Faulkner, 2009;Shipton et al., 2002). Outcrop studies have shown flow channelling and emphasise the strong spatial and temporal heterogeneity of fault zones (Bond et al., 2017;Burnside et al., 2013;Dockrill and Shipton, 2010;Eichhubl et al., 2009;Schulz and Evans, 1998;Soden et al., 2014). If fracture networks or faults are close to failure due to tectonically induced changes in the stress conditions or changes in pore pressure, vertical fluid flow is enhanced (Barton et al., 1995;Wiprut and Zoback, 2000). This so called fault-valve behaviour, where 55 faults act as highly permeable pathways for fluid discharge, is particularly likely for faults that remain active while unfavourably oriented for reactivation within the prevailing stress field (Sibson, 1990). Geomechanical parameters such as slip tendency (Morris et al., 1996) or fracture stability (Handin et al., 1963;Terzaghi, 1923) can be used to assess the potential of vertical fluid flow. The latter considers pore pressure which is a critical parameter controlling reservoir integrity not only with regards to fault weakening (Hickman et al., 1995) but also with respect to the integrity of the caprock (Caillet, 1993;Sibson, 60 2003). The need for improved understanding of fracture networks and the potential of fracture reactivation and/or hydromechanically fracturing of caprock due to the injection of CO2 has been highlighted by experiences at existing industrial CO2 storage projects. At the Sleipner storage site, fractures in thin caprock layers appear to control the size and extent of the CO2 plume (Cavanagh et al., 2015). The storage site of In Salah, Algeria, where between 2004 and 2011 around 4 million tons of CO2 were injected 65 into an anticlinal structure at ~1,800 m depth, has been the focus of many studies on fracture reactivation and hydraulic fracturing of caprocks as observations at the end of the injection period suggested that pressure had migrated vertically into the caprock (Bond et al., 2013;Michael et al., 2010;Rutqvist et al., 2010;Stork et al., 2015). The existing data indicate that injection pressures were too high for the low permeability reservoir rock and hydraulically fractured the reservoir and the lower caprock, potentially also reactivating pre-existing fracture networks related to small scale faults (White et al., 2014). 70 To study how vertical fluid flow along fault zones may be related to geomechanical parameters we examine the naturally occurring CO2 reservoir of the St. Johns Dome, located at the border of Arizona and New Mexico. At this site, migration of fluids from the subsurface reservoir to the surface is directly linked to faults which extend through the caprock (Gilfillan et al., 2011;Moore et al., 2005), with leakage having occurred for at least 420 ka and is still ongoing (Miocic et al., 2019a;Priewisch et al., 2014). We show that leakage locations are controlled by the orientation of the reservoir bounding fault with respect to 75 the regional stress field. Geological setting The St. Johns Dome (or Springerville-St. Johns Dome) natural CO2 reservoir has more than 4.7 x 10 10 m³ of recoverable CO2 and is located on the southeastern edge of the Little Colorado River Basin on the Colorado Plateau ( Fig. 1) near to the Transition Zone between the Basin and Range and Rio Grande Rift tectonic provinces (Bashir et al., 2011;Rauzi, 1999). It is one of 80 sixteen known naturally occurring CO2 reservoirs on the Colorado Plateau and one of the few known naturally occurring CO2 reservoirs world-wide where fluids are leaking to the surface (Gilfillan et al., 2008(Gilfillan et al., , 2009Miocic et al., 2016). The CO2 reservoir lies within a broad, NW-trending anticline that is intersected by the steeply dipping NW-SE trending Coyote Wash fault ( Fig. 2, Moore et al., 2005;Rauzi, 1999). This major fault appears to also to form the western boundary of the productive portion of the former commercially exploited St. Johns Dome CO2 gas field. Normal displacement across the fault ranges from 85 less than 30 m (Salado Springs) to more than 200 m at the apex of the Cedar Mesa Anticline, 25 km SE of Salado Springs (Embid, 2009). The fault is thought to be related to Paleogene Laramide compressional tectonics which led to monoclinal folding of the Phanerozoic strata and the reactivation of older basement structures such as the Coyote Wash Fault on the Colorado Plateau (Marshak et al., 2000). The normal displacement of the fault suggests an inversion of the reverse fault related to the Basin and Range extension starting in the Early Miocene and continuing in the Pliocene as evident from displacement 90 of Pliocene basalt flows (Embid, 2009). The Permian reservoir rocks (siltstones, sandstones and limestones) which discordantly overlie Precambrian granites (Fig. 3) are relatively shallow at 400-700 m depth and the CO2 is present in the gas state (Gilfillan et al., 2011). Anhydrite and mudstone beds within the Permian rocks divide the reservoir vertically into several producing zones while Triassic and Cretaceous calcareous shales and mudstones act as seals (Fig. 3). The Permian strata include, from oldest to youngest, the Supai Formation, which consist of the Amos Wash Member, Big A Butte Member, Fort Apache 95 Member, Corduroy Member, and the San Andres Limestone Glorieta Sandstone. A detailed geological description of the Permian Rocks can be found in Rauzi (1999). The current gas-water-contact is at 1425 m above sea level and the reservoir not filled-to-spill. The surface rocks are mainly Triassic to Quaternary sediments, Plio-Pleistocene volcanic rocks and travertine deposits (Fig. 2). To the NW the CO2 reservoir is bordered by the Holbrook Basin (Harris, 2002;Rauzi, 2000) and it is closely associated with the Plioۖ -Pleistocene Springerville volcanic field which lies just to the south and south-west of the CO2 reservoir 100 (Crumpler et al., 1994;Sirrine, 1958). The basaltic volcanic field consists of more than 400 individual vents and related flows, with the oldest volcanic activity dating back to around 9 Ma and the youngest flows, which can be found 8 km NW of Springerville, to about 0.3 Ma (Condit et al., 1993;Condit and Connor, 1996). As the CO2 within the reservoir is of magmatic origin (Gilfillan et al., 2008(Gilfillan et al., , 2009(Gilfillan et al., , 2011, charging of the reservoir is thought to be the result of degassing of magma underneath the volcanic field, with CO2 migrating along fractures and faults through the basement into the reservoir (Miocic et al., 2019a). 105 Expression and timing of fluid flow The travertine deposits at St. Johns Dome are an expression of CO2-charged fluids migrating from the subsurface to the surface. 120 Travertine formation occurs when CO2-rich fluids outgas CO2 as they migrate upwards to shallower depths and lower pressure, resulting in CaCO3 supersaturation and carbonate precipitation. As such, the St. Johns Dome travertine deposits cover a surface area of more than 30 km 2 , spread out over more than 300 km² (Figs. 2 & 4), making them one of the greatest concentrations of travertine deposits in North America. Spatially, the travertine deposits are particularly concentrated in a 10 km long zone between Salado Springs and Lyman Lake (Fig. 4, Gilfillan et al., 2011;Moore et al., 2005). This area, where present day 125 travertine formation occurs (Priewisch et al., 2014), is bounded by the buried Coyote Wash Fault and the distribution of the travertine deposits and active springs suggests that the local groundwater hydrology has been influenced by the Coyote Wash Fault (Embid, 2009). Analyses of surface springs, groundwater wells and CO2 wells with respect to the CO2 composition, water composition and noble gas concentrations have shown that samples taken along the Coyote Wash Fault trace are influenced by waters from depth that have been enriched in mantle derived 3 He and Ca (Gilfillan et al., 2014(Gilfillan et al., , 2011Moore et 130 al., 2005). Several modelling approaches emphasise the importance of the Coyote Wash Fault for CO2 and He migration from the Supai Formation to the surface (Allis et al., 2004;Keating et al., 2014) as in all models migration of gas to the surface occurs only if the fault forms a permeable conduit through the cap rocks. Soil-flux measurements indicate that there is no diffuse CO2 leakage through the cap rocks, suggesting instead that faults have controlled localized fluid flow . In addition to the occurrences along the NE tip of the Coyote Wash Fault (cluster A), travertine mounds follow the trace 135 of the Buttes Fault, of which the subsurface extent is not well constrained, over a distance of more than 7 km (cluster B). Travertine mounds are also found NE of the present-day extent of the CO2 reservoir, with no clear link to other structural elements (cluster C). It is notable that there are no indications for fluid migration in the southern half of the reservoir. U-series dating of the travertine mounds shows that leakage of CO2 from the reservoir to the surface has occurred for at least 420 ka (Fig. 4, Miocic et al., 2019a;Priewisch et al., 2014). Several of the samples analysed by Miocic et al. (2019a) fall outside the dating limitations of the U-Th method (~500 ka), which indicates that leakage may have occurred over much longer time-scales. This is not surprising given the age of the Springerville volcanic field (earliest activity ~9 Ma) from where the 150 magmatic CO2 is almost certainly sourced. Individual seeps along the Buttes Fault have lifespans of up to 200 ka and the massive travertine platform between Salado Springs and Lyman lake has at least a similar lifetime. Volumetric calculations indicate that the subsurface reservoir is constantly or regularly recharged, as several times the volume of CO2 stored in the current reservoir has leaked in the past (Miocic et al., 2019a). However, due to the long timeframe of leakage recorded by the travertine deposits, only a very low percentage (0.1-0.001 %) of the reservoir volume (1900 Mt CO2) has leaked annually and 155 thus the site could still be seen as a suitable carbon storage site from a climate mitigation point of view (Miocic et al., 2019a). These observations illustrate that fluid migration at the St. Johns Dome occurs along fault zones and once migration pathways have been established they are spatially fixed for long periods (>100 ka). This is in contrast to other fault-controlled fluid migration pathways on the Colorado Plateau, for which it is suggested that these stay open only episodically for few thousand years after rapid fault movement and subsequently heal (Frery et al., 2015). Similar cyclic reopening and healing of fractures 160 governing fault zone permeability has been recorded by travertine deposition at other active fault zones in Italy (Brogi et al., 2010). Spatially and temporally fixed migration pathways are concerning for subsurface storage sites and thus the processes controlling vertical fault zone permeability at the St. Johns Dome are analysed herein. Methods In order to investigate the mechanisms governing the vertical fluid flow at the St. Johns Dome, a geomechanical analysis of 165 the Coyote Wash Fault was conducted using slip tendency and fracture stability approaches. Slip tendency (Ts) is a method that allows for a fast assessment of the tendency of a surface to undergo slip under the present-state effective stress field. It is the ratio of resolved shear stress to resolved normal stress on a surface (Morris et al., 1996) where τ is the shear stress and σn the effective normal stress acting on the fault. Slip is likely to occur on a surface if Ts≥ µs 170 with µs being the coefficient of static friction which is generally assumed to be 0.6 (Byerlee, 1978;Moeck et al., 2009;Sibson, 2003). Fracture stability (Fs) is the increase in pore pressure that is required to reduce the effective stresses such that the fault plane is forced into failure (Handin et al., 1963;Terzaghi, 1923). In contrast to slip tendency, Fs takes rock properties such as cohesion and angle of internal friction into account and thus fault rock composition needs to be known. A 3D geological model of the St. Johns Dome was built based on published geological maps (Embid, 2009;Sirrine, 1958), 175 well data from 37 exploration and production wells available from the Arizona oil and gas conservation commission (well logs, horizon markers) and previously published reservoir horizon map and markers (Rauzi, 1999) using Move™. Between wells a constant stratigraphic thickness was assumed and for the fault a dip of 70° was estimated, based on previous works (Embid, 2009, Rauzi, 1999 and a 3D dip-domain construction (Fernandez et al., 2008) of the intersection of the fault trace with the 1/3 arc-second DEM of the 3D elevation programme of the USGS. The modelled fault has 6635 faces constructed as 180 triangles from 3525 vertices. Cut-off-lines were created on the fault surface by extracting the dip from a 200 m wide patch of the horizon of interest on either side of the fault and projecting this along the dip-direction until it intersects with the fault (Yielding and Freeman, 2016). The current gas-water-contact is at 1494 m above sea level (Rauzi, 1999) and is assumed to be horizontal. Due to lack of pressure data, a hydrostatic pressure gradient is assumed (0.0105 MPa/m). Geomechanical analysis of the model was conducted with industry standard software (Move™ and TrapTester®). As no outcropping fault rocks were 185 available, the Shale Gouge Ratio (Yielding et al., 1997) was used as a fault rock proxy. SGR was calculated from a Vshale log of well 10-29-31, which was calculated from the gamma ray log assuming a linear response (Asquith and Krygowski, 2004). As this method only applies to siliciclastic rocks, zonal Vshale values for evaporitic sequences (70% shale content for anhydrite, 55% shale for carbonates were assumed, expecting rapid fault sealing for these lithologies (Pluymakers and Spiers, 2014) or low permeability fault rocks (Michie et al., 2018)) were used. Resulting SGR values indicate a high potential of phyllosilicate 190 rich fault rocks (Fig. 5). To emphasise the uncertainty regarding the fault rock composition, two different fault rocks were used for Fs calculations: clay smear (cohesion C=0.5 MPa, coefficient of internal friction µ=0.45) and phyllosilicate (C=0,5, µ=0.6) with rock strength values from the TrapTester® internal database. Note that for modelling purposes we assume a siliciclastic sequence, however the stratigraphic sequence also contains ~15 % carbonate and evaporitic rocks (Fig. 3) which may have locally significant influence on the fault rock strength. Ts results are presented using stereonets as this here the reader can 195 readily visualise how changes in the stress field orientation would influence fault stability while Fs results are presented on a Mohr circle as this allows a direct visualisation of how much the pore pressure needs to change to force different parts of the fault into failure. It also allows the reader to see how changes in fault rock strength could change the pore pressure needed for fault failure. For stress field no in situ stress measurements were available, however in addition to World Stress Map data (Heidbach et al., 2016) the nearby Springerville volcanic field can be used to derive the orientation of the horizontal stresses 200 as presented in the following. Stress field at the St. Johns Dome The location of the St. Johns Dome reservoir at the margin of the Colorado Plateau and within the greater Basin and Range province has significant impact on the stress field in the study area. It is clear that the regional stress field is highly variable as shown by the available stress field data in the vicinity of the St. Johns Dome (50 km radius, Fig. 2) from the World Stress Map 210 (Heidbach et al., 2016) combined with a regional study on volcanic vent orientation in the Springerville volcanic field (Table 1, Connor et al., 1992). Note that the maximum horizontal stress (SHmax) from Connor et al. (1992) is based on vent clusters linearly aligned with lengths of 11 to 20 km length (Fig. 2) and that table 1 lists them as point measurements at the centre of the cluster. To the south and south-east of the CO2 field, the SHmax is oriented NE-SW while west of the reservoir the SHmax orientation is highly variable, ranging from NW-SE to E-W (Fig. 2). While these datapoints are associated with an uncertainty 215 of at least ±15°, the orientation of the stress field for the St. Johns Dome faults is difficult to constrain. A normal faulting regime [vertical stress (Sv) > maximum horizontal stress (SHmax) > minimum horizontal stress (Shmin)] is assumed as based on the World Stress Map and works by Kreemer et al. (2010) and Wong and Humphrey (1989) for this area of the Colorado Plateau. Integration of density logs (wells 10-29-31 and 11-16-30) gives a magnitude of Sv of 23 MPa/km. Minimal horizontal stress in normal faulting regimes is typically about 65-85% of the vertical stress (Hillis, 2003), which gives a magnitude of 220 Shmin in the range of 15 to 19.5 MPa with the magnitude of SHmax set between Sv and Shmin. As the reported stress field measurements appear to form three clusters (Tab. 1), three different stress fields were defined (Tab. 2): stress field A is similar to the stress fields indicated by measurements 7 and 8, with SHmax having an azimuth of 140°; stress field B is oriented similar to the stress field measurements 5 and 6 with a SHmax azimuth of 100°; and stress field C is similar to the stress fields indicated by measurements 2 to 4 with a SHmax azimuth of 50°. The solitary north-south stress field 225 measurement (ID=1) was not considered further. For the ~NW-SE trending Coyote Wash Fault these stress fields also represent the most-(A), moderately-(B) and least-likely (C) cases for fault reactivation. Geomechanical analysis was conducted under all three defined stress fields. Geomechanical controls on vertical fluid migration The results of the geomechanical analysis of the Coyote Wash Fault highlight that the orientation of the stress field has a major 235 impact on both the slip tendency (Fig. 6) and fracture stability (Fig. 7). Slip tendency indicates that for stress field A most parts of the fault are close to failure (Ts>0.5), for stress field B the fault is only intermediately stressed (0.3<Ts<0.5), and for stress field C the fault is far away from failure (Ts<0.2). Similarly, only slight increases of pore pressure are needed to force the fault into failure under stress fields A and B and a clay smear fault rock (0.95 MPa and 1.33 MPa, respectively). The pore pressure increase needed to force the fault into failure in case of stress field C is much higher at 6.21 MPa. Note that slip tendency in 240 both stress fields A and B is higher at the NW tip of the fault than in the SE section of the fault (Fig. 6), indicating that failure is more likely to occur in the NW. This is also true for the spatial distribution of fracture stability which for stress fields A and C (most and least likely to fail) and a clay smear fault rock is illustrated in Figure 7. The results of the geomechanical analysis show that the bounding fault of the St. Johns Dome CO2 reservoir is intermediately to critically stressed for two of the three modelled stress fields (A and B). For the same stress fields a weak fault rock within 245 the Coyote Wash Fault zone results in fracture stabilities which range from less than 1 MPa to 1.33 MPa. The most critically stressed areas are located at the NW tip of the fault (Salado Springs) while the SE part of the fault is relatively stable for all studied stress fields. The Fs values of 0.95 MPa and 1.33 MPa for a clay smear fault rock translate to an additional CO2 column of ~110 m and ~160 m, respectively. Currently the reservoir is not filled-to-spill and the 3D geological model indicates that the reservoir 250 interval at the NW part of the fault could retain an additional ~150 m of CO2 column. Thus, additional filling of the reservoir with a third to half more CO2 by volume could lead to fault failure and vertical fluid migration along the fault. Evidence that the reservoir has held more CO2 in the past is provided by older travertine deposits located outside the present day extent of the subsurface reservoir (Figs. 2 & 3, Miocic et al., 2019a) and the fact that higher paleo-reservoir pressures have been implied by a geochemical study . These higher reservoir pressures were likely enough to bring the NW part of the 255 fault close to failure and we suggest that the permeability of fracture networks within the critically stressed fault damage zone was therefore increased (Barton et al., 1995;Ito and Zoback, 2000;Min et al., 2004). In order to sustain the long periods of leakage recorded by the spatially stable travertine deposition the fault must be critically stressed for similarly long periods. Indeed, volume calculations on how much CO2 must have leaked to the surface based on the travertine deposits show that one to two orders of magnitude more CO2 was lost from the reservoir than it can hold (Miocic et al., 2019a). It is suggested that 260 the continuous influx of magmatic CO2 degassing from beneath the Springerville Volcanic Field into the reservoir caused the fault to be close to being critically stresseda reasoning also supported by this study. The geomechanical analysis also demonstrates that a change of the fault orientation within the stress field should not be underestimated and can lead to failure along one part of a fault while large parts of the fault are geomechanically stable. The strike direction of the Coyote Wash Fault changes from ESE-WNW in the southern part of the fault to NW-SE in the northern section and this change in strike is enough to render the northern section critically stressed (in two of the stress fields modelled; 270 A & B) -with leakage pathways being the result (Fig. 9). Higher paleo gas columns within the reservoir likely contributed the forcing the fault into failure at the northern fault tip. However, some sections of the fault in the SE also have relatively low fracture stability values (Fig. 8A) which translate to only 10's of meters more supported gas column than the NW section. Yet, there are no indications for past or present leakage in the SE part of the St. Johns Dome. We argue that the stress field orientation in the SE is different from the stress field orientation in the NW area of the St. Johns Dome and that as a result the 275 fault is far from failure towards its SE tip. This is supported by stress field measurements in the vicinity of the southern edge of the reservoir (Fig. 2, Tab. 1) which imply a NE-SW SHmax orientation. Vertical migration of fluids through fault and fracture networks or corridors can be classified by their location in (1) the fault damage zone, (2) at the fault tip, and (3) at the crest of a fold (Ogata et al., 2014). As evidenced by travertine deposits vertical fluid migration at the St. Johns Dome occurred at all three types of fracture networks (Fig. 2), but considerably larger volumes 280 of fluid migrated through fracture networks linked to faults, particular at their NW tips. This indicates that, at least at this site, faults are a higher risk factor for leakage than other migration pathways such as fracture networks along the anticline structure or capillary leakage though a caprock. Based on travertine volumes the largest volumes of leakage occurred at the NW tip of the Coyote Wash Fault, in the area between Lyman Lake and Salado Springs (Figs. 2, 8, Miocic et al., 2019a). This indicates high permeability fracture networks within the damage zone close to the fault tip as predicted by numerical models (Backers 285 and Moeck, 2015;Zhang et al., 2008). The lack of similar leakage pathways observed at the SE tip of the fault can be attributed to the different stress field orientation. For geological storage in general the occurrence of large-scale leakage at fault tips is also concerning as displacement at fault tips usually is low and as such fault tips are not seismically resolvable and may remain undetected. Thus seismically resolved faults could be extended beyond the normally picked extend to include the fault tips. Similarly, faults with low displacement such as the Buttes Fault, for which significant fault related leakage has been recorded 290 but is thought to have a maximum displacement of <25 m, may not be detectable on seismic data. This highlights the need for a good structural understanding of any geological storage site to ensure that fault tips and small faults are considered and incorporated, possibly as an additional uncertainty parameter, into the geological model. 310 While the geomechanical analysis highlights the role of critically stressed faults for fluid migration at the St. Johns Dome, it is missing in situ stress field data from within the CO2 reservoir. Such data is crucial for a detailed and reliable study of fracture and fault stability (e.g., Becker et al., 2019), however there are cases where such in-situ data is missing and a geomechanical analysis may be needed (Henk, 2005). In particular during the site selection and appraisal of subsurface storage sites a preliminary geomechanical analysis based on existing stress field data can identify potentially critically stressed faults. The 315 lack of in-situ data can be compensated by studying several plausible stress fields (as in this study) and including uncertainties into the geomechanical analysis (Ziegler and Heidbach, 2020). For the latter, uncertainties in the stress field orientation and magnitude and in the fault orientation should be included. Statistical approaches such as Bayesian or Markov Chain Monte Carlo modelling can be useful to identify uncertainty thresholds and to determine the precision by which the geomechanical parameters need to be known in order to have reliable fault and fracture stability predictions (Bao et al., 2013;Chiaramonte et 320 al., 2008;McFarland et al., 2012). For the geomechanical prediction of permeable fracture networks and thus leakage pathways at the St. Johns Dome the stress field orientation is integral. The location of the natural CO2 reservoir at the edge of the Colorado Plateau is the likely reason for the stress field orientation change, with clear changes in crustal composition and strength in the vicinity of the St. Johns Dome (Hendricks and Plescia, 1991;Qashqai et al., 2016). The study area is also located at the intersection of the NE-NNE 325 trending Jemez Lineament, a tectonically active zone that is characterized by Paleogene-Quaternary extension and volcanism ( Fig. 1), and the ESE trending Arizona Transition Zone (Aldrich and Laughlin, 1984;Kreemer et al., 2010). Additionally, the presence of salt deposits in the Holbrook Basin north of the study area may also impact the local stress field (Neal and Colpitts, 1997;Rauzi, 2000). The complex regional setting at the St. Johns Dome and the associated uncertainties for geomechanical modelling further highlight the need for thorough site selection criteria for engineered fluid storage sites and adequate 330 geological data to ensure that only reservoirs with well understood structural frameworks are chosen. Implications for geological storage applications Geomechanical modelling suggests that vertical fluid migration from the reservoir to the surface at the St. Johns Dome natural CO2 reservoir is controlled by fracture networks in the damage zone and tip of near-critically stressed faults. We propose that regular filling of the reservoir with CO2 from mantle sources increased the pore pressure within the reservoir and further 335 reduced the stability of near critically stressed faults, leading to the leakage of large volumes of CO2 over the time-span of several 100 kas. While the leakage rates at the St. Johns Dome are low enough to render the faulted site an adequate CO2 store for climate mitigation, similar leakage rates could socially and operationally impede geological storage of methane or hydrogen, in particularly at onshore storage sites. For fault-bound subsurface storage sites for CO2 or other fluids the history of geomechanically controlled leakage at the St. 340 Johns Dome clearly illustrates the need for a good understanding of regional and local stress fields and faults. In particular the stress state of faults and fault related fracture networks prior to fluid injection needs to be well understood in order to reduce the risk of vertical fluid migration through fractured caprock. We recommended to select areas where there are no significant regional stress field changes as these complicate geomechanical predictions. Indeed, in situ stress data from wells are key for any advanced leakage risk prognosis. To further understand the leakage mechanisms at the St. Johns Dome geomechanical 345 modelling of the Buttes Fault, combined with an uncertainty assessment, is recommended. More detailed dating of the travertine deposits could reveal at which part of the faults (fault tip vs fault damage zone) failure occurred first and provide insights into the time dynamics of leakage. Author contribution JM and SG designed the research project which was carried out by JM with help from GJ and input from SG. JM prepared the 350 manuscript with contributions from all co-authors. Competing interests The authors declare that they have no conflict of interest.	2020-02-13T09:22:21.134Z	2020-02-10T00:00:00.000	{ "year": 2020, "sha1": "5ccb02d4a90a08e1e991502ce415704e858bbe7b", "oa_license": "CCBY", "oa_url": "https://se.copernicus.org/articles/11/1361/2020/se-11-1361-2020.pdf", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "4d568b97c0c3a3e17987b9637bbce3ed8bc79962", "s2fieldsofstudy": [ "Geology" ], "extfieldsofstudy": [] }
250643056	pes2o/s2orc	v3-fos-license	Gregor Mendel’s legacy in quantitative genetics Gregor Mendel’s discovery of the laws of segregation and independent assortment and his inference of the existence of non-mendelian interactions between loci remain at the heart of today’s explorations of the genetic architecture of quantitative traits. The field of genetics was born with the publication in 1866 of Gregor Mendel's Experiments in Plant Hybridization [1]. Working with the garden pea, Pisum sativum, Mendel chose 7 "characters" (polymorphic loci each affecting a different phenotype in today's parlance) that "permit of a sharp and certain separation," excluding those for which "the difference is of a 'more or less' nature, which is often difficult to define" [1]. Mendel chose these loci because the hybrid (F 1 ) between the homozygous parental genotypes (to use modern terminology) was indistinguishable from one of the parents; i.e., one of the alleles was dominant and the other recessive (Mendel's original terms). He excluded phenotypes that were intermediate between the 2 parents in the F 1 hybrid. When Mendel crossed each of the 7 F 1 hybrids among themselves, he observed a 3:1 ratio of dominant to recessive phenotypes in the F 2 generation. He inferred that the F 1 hybrids have 2 alleles at each locus (A, a), which are present in equal proportions in the female and male gametes, and combine at random to produce F 2 offspring in the ratio 1 AA:2 Aa:1 aa (the Law of Segregation). Mendel also performed crosses in which the hybrids differed for 2 or 3 of the 7 loci, analyses of which led him to conclude "the relation of each pair of different characters in hybrid union is independent of the other differences in the 2 original parental stocks" (the Law of Independent Assortment) [1]. Luckily for Mendel, the 7 loci were each on a different autosome. In a separate series of crosses between 2 species of common bean with different flower colors and unexpected ratios of flower color in hybrids, Mendel correctly inferred multiple loci with recessive epistasis (where the expression of one gene is modified by another). Mendel's principles of inheritance were contrary to the common observation at the time that crosses between organisms with different phenotypes were often intermediate between the 2 parents, and that phenotypic variation in populations is continuous and not discrete [2]. Indeed, Mendel deliberately excluded such phenotypes (or traits) from consideration. These continuously varying phenotypes, now recognized as quantitative traits, are more common in populations than phenotypes with the inheritance properties that Mendel reported. The relationships between relatives for quantitative traits are described in terms of regressions and correlations, as defined by the early biometricians Galton and Pearson [3]. It that Ronald Fisher reconciled the 2 viewpoints [4] by showing that mendelian inheritance at a large (essentially infinite) number of loci would give rise to the observed continuous variation by generalizing Mendel's principles to alleles with small effects, any type of dominance or epistasis, nongenetic (environmental) effects, and random mating populations (Fig 1). In the absence of knowledge of the individual genes and causal alleles, statistical models based on correlations of phenotypes between relatives are used to determine the fraction of variation for a quantitative trait in a population that is attributable to genetic variation, and to predict the response to selection. Quantitative traits include all aspects of an organism's morphology, physiology, behavior, and fitness, as well as molecular phenotypes such as gene expression, and protein and metabolite abundances. Although the principles of mapping regions of DNA associated with [1] at a single locus for any phenotypic difference and any degree of dominance [4]. He centered the difference in phenotypes between the 2 homozygous genotypes at 0, so the genotype increasing the trait value has an effect of +a and the genotype decreasing the trait value has an effect of −a, where a is the additive effect. He defined the effect of the heterozygote, d (the dominance effect), as the difference between the average phenotype of heterozygotes and the average of the 2 homozygotes. (b) Fisher then assumed that each locus affecting a quantitative trait segregates in natural populations, with the frequencies p and q (q = 1-p) for the A 1 and A 2 alleles, respectively. The frequencies of the genotypes or phenotypes are given by the binomial expansion of (p+q) n-1 , where n is the number of genotypes or phenotypes. With random mating, the frequencies for genotypes A 1 A 1 , A 1 A 2 , and A 2 A 2 at each locus are p 2 , 2pq, and q 2 , respectively. In this example, a = 5, d = 0, and p = q = 0.5, so the 3 genotypes have phenotypes of 5, 0, and −5 with frequencies of 0.25, 0.5, and 0.25. These values of p and q correspond to the frequency of the 2 alleles in Mendel's crosses of F 1 hybrids, with the expected genotype ratios of 1:2:1 in the F 2 , but are generalized to any phenotypic effects and allele frequencies. (c) Fisher then assumed that as more loci affect the trait, the phenotypic range remains the same, and that the effects of all loci are the same or nearly so and add together to give the observed phenotype. Thus, as the number of loci increases, the effects of each on the phenotype become smaller. This example is for 2 loci (A and B), each with 2 alleles, with the same values of a, d, p, and q for each locus as that for locus A in panel (b). There are 9 genotypes but only 5 phenotypes. (d) The distinction between phenotypes decreases even further as the number of loci increases, and nongenetic environmental effects result in truly continuous distributions of quantitative traits in natural populations. In this example, there are 5 loci (A, B, C, D, and E), each with 2 alleles, with the same values of a, d, p, and q for each locus as that for locus A in panel (b). There are 243 genotypes but only 11 phenotypes. https://doi.org/10.1371/journal.pbio.3001692.g001 particular traits, known as quantitative trait loci (QTLs), by linkage to mendelian loci have been understood since the beginning of the 20th century [5], these studies were not widespread until the discovery of polymorphic molecular markers with strictly mendelian inheritance and the development of sophisticated statistical QTL mapping methods in the late 20th century [6]. Prior to that, mendelian and quantitative genetics were separate fields, with forward genetic screens for mendelian mutations in model organisms being used to determine the genetic basis of development, physiology and behavior, and quantitative genetic analysis of naturally occurring variation being used for genetic improvement of agricultural species and evolutionary biology. What have we learned about the genetic basis of variation for quantitative traits from the thousands of studies mapping QTLs that have now been completed? First, quantitative traits are indeed highly polygenic, as suggested by Fisher [4]. Second, QTL effects estimated by averaging over different genetic backgrounds in the population are small, also consistent with Fisher's hypothesis [4]. However, some observations hint that the true genetic basis of variation for quantitative traits is more complicated than there just being many loci with segregating alleles of small effects. When we introduce alleles associated with variation in a quantitative trait into common genetic backgrounds, we often find that the alleles have large effects that vary according to the genetic background [7]. Crossing mutations with large mendelian effects onto different naturally occurring genetic backgrounds in model organisms gives rise to a wide range of outcomes in which the mutant phenotype is exacerbated or ameliorated, revealing the existence of naturally segregating alleles that interact epistatically with the mutant allele. In humans, rare mendelian diseases typically have variable severity of symptoms, age of onset, and disease progression for different children with the same mutation, even within a family; again indicating the presence of naturally occurring epistatic modifiers. QTL mapping in model organisms has shown that QTL effects often depend on environmental context and sex [8]. Pervasive context-dependent effects mean that small effects of QTLs in populations may reflect averaging of true effects across multiple contexts. These features of the genetic architecture of quantitative traits have implications for agriculture, human genetics, and evolution. For example, unless the relevant contexts are accounted for, attempts to improve traits of agricultural importance by transferring QTLs from breeds and varieties selected for performance under one set of conditions to breeds and varieties adapted to different conditions may not have the expected effects, and genomic prediction developed for one breed may not transfer to other breeds. Polygenic risk scores for human diseases that have been developed for one population may not be accurate in other populations unless specific interactions are included in the models. Identifying epistatic modifiers of rare human diseases could provide clues for therapies, and defining genotypes by their drug environment interactions will facilitate pharmacogenomic applications. Furthermore, contextdependent effects in natural populations may be in part responsible for the maintenance of quantitative genetic variation and adaptive evolution. Mendel's experiments on plant hybridization have laid the foundation of modern quantitative genetics. Mendel was well aware of the significance of his discoveries [9]. A few months before his death, during the investiture of his successor Franz Barina as abbot, he reportedly said "My scientific work has brought me great joy and satisfaction; and I am convinced that it won't take long that the entire world will appreciate the results and meaning of my work" [10]. Although the accuracy of this quote cannot be unambiguously verified, Mendel would have been right. But little could he have predicted the enormous impact of his discoveries for agriculture, evolutionary biology, and precision medicine more than 150 years after publication of his treatise and 200 years after his birth.	2022-07-20T06:17:37.783Z	2022-07-01T00:00:00.000	{ "year": 2022, "sha1": "114fef98290841ed4362968b6c455ffc07922811", "oa_license": "CCBY", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "de38a847b9f8c56c644c0954f30b772d6cdd89d1", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
6995827	pes2o/s2orc	v3-fos-license	Prospects for gene-engineered T cell immunotherapy for solid cancers Adoptive transfer of receptor-engineered T cells has produced impressive results in treating patients with B cell leukemias and lymphomas. This success has captured public imagination and driven academic and industrial researchers to develop similar 'off-the-shelf' receptors targeting shared antigens on epithelial cancers, the leading cause of cancer-related deaths. However, the successful treatment of large numbers of people with solid cancers using this strategy is unlikely to be straightforward. Receptor-engineered T cells have the potential to cause lethal toxicity from on-target recognition of normal tissues, and there is a paucity of truly tumor-specific antigens shared across tumor types. Here we offer our perspective on how expanding the use of genetically redirected T cells to treat the majority of patients with solid cancers will require major technical, manufacturing and regulatory innovations centered around the development of autologous gene therapies targeting private somatic mutations. p e r s p e c t i v e Adoptive transfer of receptor-engineered T cells has produced impressive results in treating patients with B cell leukemias and lymphomas. This success has captured public imagination and driven academic and industrial researchers to develop similar 'off-the-shelf' receptors targeting shared antigens on epithelial cancers, the leading cause of cancer-related deaths. However, the successful treatment of large numbers of people with solid cancers using this strategy is unlikely to be straightforward. Receptor-engineered T cells have the potential to cause lethal toxicity from on-target recognition of normal tissues, and there is a paucity of truly tumor-specific antigens shared across tumor types. Here we offer our perspective on how expanding the use of genetically redirected T cells to treat the majority of patients with solid cancers will require major technical, manufacturing and regulatory innovations centered around the development of autologous gene therapies targeting private somatic mutations. Irrefutable evidence that an entirely immunologic approach can cause regression of a wide array of human cancers has come from the recent success of using monoclonal antibodies (mAbs) targeting checkpoints of immune activation, including cytotoxic T lymphocyte-associated protein 4 (CTLA-4) (ref. 1) and programmed cell death protein 1 (PD-1) (ref. 2). This includes patients affected with an ever-expanding list of malignancies, including melanoma 1,2 , renal cell carcinoma 2,3 , lung cancer 2,4 , bladder cancer 5 , ovarian cancer 6 , Hodgkin's lymphoma 7 , and gastrointestinal (GI) and endometrial cancers associated with defects in DNA mismatch repair 8 . Despite different mechanisms of action, these immunotherapies culminate with the activation and expansion of tumor-reactive T cells [9][10][11][12] . Because T cells are often are the final effectors of immune-mediated cancer regression, strategies that directly use tumor-reactive T cells as a therapy have been developed 13 . In this approach, termed adoptive cell transfer (ACT), T cells are expanded outside the potentially immunosuppressive environment of a tumor and re-infused in large numbers into the cancer patient (up to 10 11 cells). Historically, procuring antitumor T cells for use in ACT has come from the surgical removal of a cancer metastasis in order to obtain tumor-infiltrating lymphocytes (TILs). TILs demonstrate tumor reactivity with variable frequency in a range of cancers, including melanoma [14][15][16][17] , GI 18,19 , lung 20 and human papilloma virus-associated malignancies 21 . TIL infusion can induce durable complete responses (CRs) 14,21 , including in patients for whom other immunotherapies have failed 14 . Despite demonstrable efficacy, use of TIL outside the context of clinical trials performed at academic medical centers has proven challenging. Progress in gene engineering technologies has simplified the generation of antitumor T cells, overcoming many of the practical barriers that have limited wide dissemination of ACT using TIL cells. Gene engineering obviates the requirement for surgery because T cells can be isolated from the blood and receptors conveying specificity for tumor-associated antigens can be introduced using viral and non-viral integration techniques 22 . Thus, antitumor T cells can potentially be made on a large scale using commercial production methods. Indeed, recent experience with sipuleucel-T, a gene-modified cell product for prostate cancer, demonstrated the feasibility of having a patient's immune cells collected, sent to a central manufacturing facility, and returned back for re-infusion in a manner that gained US Food and Drug Administration (FDA) regulatory approval 23 . Finally, genetic modification of T cells has a track record of safety. Gammaretroviral and lentiviral vectors have been used most commonly in antigen receptor gene therapy trials. Despite concerns about the possibility of insertional mutagenesis 24 , introduction of antigen receptors into mature human T cells has been used to treat several hundred patients without evidence of clonal expansion or transformation 25 . Collectively, a framework of manufacturing feasibility, regulatory precedent and vector safety is now in place and it is possible to envision treating large numbers of cancer patients using gene-engineered T cells. Recent success with gene-modified T cells targeting the B cell lineage differentiation antigen CD19 in a range of B cell malignancies has focused attention on using similar 'off-the-shelf ' antigen receptors to treat patients with advanced solid cancers. In this Perspective, we offer our appraisal of how adoptive immunotherapy using receptor-engineered T cells can enter mainstream clinical oncology for patients with advanced epithelial cancers, the leading cause of cancer-related deaths 26 . Antigen receptor-engineered T cells T cell receptors. Genetically redirecting a T cell's specificity toward a patient's cancer can be accomplished by the introduction of one of two types of antigen receptors. In one approach, a cloned T cell receptor (TCR) conferring tumor recognition is inserted into circulating lymphocytes. Similarly to the endogenous TCR expressed by all T cells, genetically introduced TCRs recognize a proteolytically processed peptide derived from either a cytosolic or membrane-associated Clinical trials. Since 1994, at least 148 human clinical trials have been initiated in the US for testing gene-engineered TCRs or CARs for the treatment of cancer (Fig. 1a) (ref. 35). Additional trials are also being conducted in Europe, Asia and Australia. Although the number of new TCR trials has remained relatively constant in recent years, the number of trials evaluating CAR-modified T cells has grown exponentially. In the majority of cases, the antigens targeted in these trials have been shared by tumors and normal tissues (Fig. 1b). Results from these trials have demonstrated that receptor-engineered T cells can, in some cases, mediate long-term remissions of selected solid and hematologic cancers. For example, a recently completed TCR trial 36 targeted a human leukocyte antigen (HLA)-A2-restricted epitope derived from NY-ESO-1, a cancer germline antigen (CGA) located on the X chromosome 37 . In this study, 11 of 20 patients (55%) with metastatic melanoma had objective evidence of cancer regression, including four CRs 36 . Importantly, three of these CRs were ongoing after >36 months. An additional 18 patients with synovial cell sarcoma, an aggressive connective tissue cancer associated with a characteristic t(X;18) chromosomal translocation 37 , were treated in this same study. Eleven of 18 (61%) treated patients had an objective response, including one CR. Although NY-ESO-1 is expressed by germ cells such as the testis, these tissues do not express MHC. Consequently, no on-target but off-tumor toxicities were observed. A multitude of ACT studies have now demonstrated remarkable and frequently durable responses 38 using CARs targeting CD19, a B cell-lineage antigen expressed on the surface of both normal B cells and many malignant B cells. After the first successful application of this approach in a follicular lymphoma patient 39 , CD19-specific CARs have been used effectively to treat patients with other chemotherapyrefractory B cell cancers including marginal zone lymphoma 40 , aggressive B cell lymphomas 41 , chronic lymphocytic leukemia 38,40,42,43 , and adult and pediatric acute lymphoblastic B cell leukemias (ALLs) [44][45][46][47][48] . Additionally, a recent case report suggested that a single patient with multiple myeloma had a sustained CR after an autologous stem cell transplant administered in conjunction with CD19-specific CAR cells 49 . This finding is currently being confirmed in a larger cohort of patients. Because CD19 is expressed by normal B cells, B cell aplasia and deficiencies in circulating immunoglobulins (Igs) are frequently observed in patients receiving CD19-specific CARs; these toxicities can be managed with Ig infusions. Many patients also exhibit cytokine release syndrome (CRS), a constellation of symptoms that occur after T cell infusion, and which is attributed to an exuberant release of cytokines, such as interferon (IFN)-γ and interleukin (IL)-6 (ref. 50). Symptoms associated with CRS include fevers, hypotension, hypoxemia, cardiac dysfunction, kidney failure and electrolyte abnormalities. Some patients also develop neurologic symptoms, including expressive aphasia, tremor and seizures, the cause of which remains unknown. In the majority of cases, these side effects can be managed with aggressive supportive care alone or in combination with immunosuppressants such as steroids or cytokine-specific antibodies. Nevertheless, treatment-related deaths have occurred at many institutions. Despite these complications, given the impressive clinical responses seen in patients with otherwise recalcitrant disease, we anticipate that CD19-specific CARs will enter mainstream care for many B cell malignancies in the next 1-2 years. Limitations of current CAR and TCR approaches to treat common epithelial cancers Two principles have emerged from successful TCR-and CARengineered ACT trials to date. First, potent antitumor effects in the absence of normal tissue damage can occur if the target is uniquely expressed by a patient's tumor, as exemplified by the NY-ESO-1 TCR trials. Second, if a patient's T cells are modified with a receptor that recognizes an antigen expressed both on non-transformed tissues and cancer cells, such as CD19 in the CD19-specific CAR trials, these cells will attack and destroy both normal and malignant tissue equally vigorously. Based on the success of the NY-ESO-1 TCR and the CD19specific CARs, there is tremendous excitement in the immunotherapy field that similar 'off-the-shelf ' TCRs and CARs targeting antigens shared across tumors, such as CGAs or tissue-differentiation antigens, will be highly effective against the majority of solid cancers. Prerequisite to the generation of an antitumor CAR is knowledge of the genetic sequence from the scFv region of a mAb capable of recognizing antigens on a cancer cell's surface. Although the testing of CARs in oncology clinical trials is comparatively new 51,52 , the search for tumor-specific antibodies is not 53 . Since the initial description in 1975 of a high-efficiency method for producing mAbs 54 , there has been a massive investment by academic and industry laboratories to develop therapeutic antibodies targeting cancer cells. Compared with the size of this investment, the search has yielded relatively few results. To date, 20 mAb or mAb drug-radioisotope conjugates have been approved by the FDA for the treatment of cancer 55 . Among these, five target tumors indirectly by various mechanisms, including disruption of angiogenesis (bevacizumab, ramucirumab), interference with tumor-related bone remodeling (denosumab), or nonspecific immune activation through blockade of negative regulatory pathways such as CTLA-4 (ipilumimab) and PD-1 (nivolumab, pembrolizumab). For the remaining antibodies that directly target cancer cells, none are tumor-specific, but rather they recognize differentiation antigens also expressed by normal cells. Eight of these target lineage-specific antigens of the hematopoietic system, including CD20 (rituximab, ofatumumab, obinutuzumab, ibritumomab), CD30 (brentuximab vedotin), CD33 (gemtuzumab), CD38 (daratumumab) and CD52 (alemtuzumab) which are expressed on B cells, activated T and B cells, myeloid cells, lymphoid and myeloid cells, and lymphocytes, respectively [56][57][58][59][60] . Not unexpectedly, each of these antibodies induce cytopenias of benign cells that co-express these target antigens. Of the remaining six mAbs that directly bind structures on the surface of solid cancers, five recognize growth factor receptors that are overexpressed by tumor cells, including epidermal growth factor receptor (EGFR, also known as ERB1; cetuximab, panitumumab) and ERBB2 (also known as HER-2/neu; trastuzumab, pertuzumab, ado-trastuzumab emtansine). For these mAbs, disruption of signal transduction pathways crucial to maintaining the malignant phenotype are the major, if not the only, mechanism of action 55 . Because EGFR and ERBB2 signaling is essential in the function of keratinocytes 61 and cardiac myocytes 62 , mAbs targeting these receptors are associated with skin and cardiac toxicities. At present, the only example of an approved mAb that directly targets a solid-tumor antigen whose sole mechanism of action is immune-mediated killing is dinutuximab, which binds the glycolipid GD2 (ref. 63). Because GD2 is expressed not only by cancers, such as neuroblastoma and sarcoma, but also by peripheral sensory nerve fibers and neurons, neuropathic pain is a dose-limiting toxicity of this antibody 63 . Thus, after more than 35 years of clinical development, none of the approved mAbs that directly bind tumor cells are tumor specific, and all can mediate on-target and off-tumor toxicities. CARs for solid cancers. A similar situation appears to be the case for other tumor-targeted mAbs that are in clinical development and whose scFv regions are being incorporated into CAR designs. For example, the scFv of various antibodies targeting the shared tissue and tumor differentiation antigen mesothelin have been used to generate mesothelin-specific CARs for human clinical trials 64,65 . Mesothelin, a glycosylphosphatidyl inositol (GPI)-anchored membrane glycoprotein is highly expressed in a number of malignancies, including pleural and peritoneal mesothelioma, as well as pancreas, lung, breast, esophageal and ovarian cancer 66 . However, mesothelin is also expressed throughout the body in sensitive tissues that possess mesothelial cells. This includes the cornea, pleura, pericardium, peritoneum, tonsils, fallopian tubes and cervix. Consistent with this expression pattern, pleuritic-type chest pain was a dose-limiting toxicity in studies using mesothelin-specific antibodies conjugated to various immunotoxins 66 . Given the potential for substantial on-target toxicities, the FDA has required a protracted dose escalation in currently accruing phase 1 clinical trials testing the safety and efficacy of mesothelin-specific CARs. Recently, interim results from one mesothelin-specific CAR phase 1 study were presented in which as many as 3 × 10 7 cells were infused. Although no on-target toxicities were observed, neither was radiographic evidence of anti-tumor activity 67 . It is presently too early in the clinical development of mesothelin-specific CAR T cells to know whether mesothelin represents a viable target for gene-engineered T cells. Likewise, a recent CAR trial was initiated targeting MUC16 (ref. 68), a glycosylated mucin expressed on the surface of the majority of ovarian cancers, as well as in normal tissues that harbor mesothelial cells such as the eye 69 . As with the mesothelin-specific CAR trials, this study is also in the midst of a slow dose titration owing to safety concerns related to normal tissue targeting. When sufficient data exists to assess the safety and efficacy of CARengineered T cells targeting other antigens shared by tumors and self tissues, substantial on-target toxicities have occurred. For example, CAR-modified T cells with an scFv specific for carbonic anhydrase IX (CAIX), an enzyme expressed by some kidney cancers and normal bile duct epithelial cells, triggered liver function abnormalities and cholangitis without causing cancer regression 52 . Similarly, infusion of ERBB2-specific CAR T cells constructed using the scFv from the humanized mAb trastuzumab resulted in lethal inflammatory cytokine release in the lung 70 . This toxicity was attributed to on-target but offtumor recognition of low levels of ERBB2 expression on lung epithelial cells 71 . A more recent trial reported the administration of a separate ERBB2-specific CAR T cell in which the scFv was derived from FRP5, a mouse anti-human ERBB2-specific mAb 72 . Although no significant toxicities were reported, neither was evidence of in vivo CAR cell expansion or IFN-γ release in the blood after cell infusion. This might indicate that the engineered cells did not productively engage the target antigen. Although 1 of 19 patients treated with the FRP5derived ERRB2-specific CAR had an objective anti-tumor response, this patient received salvage chemotherapy in addition to CAR T cell infusion. Thus, it is impossible to determine the relative contributions of the infused cells versus chemotherapy. These findings contrast with experience using CD19-specific CAR T cells in which cell expansion, cytokine release, and profound antitumor and on-target immunity against normal B cells is universally observed across trials 40,41,[43][44][45][46][47][48] . A possible exception to the paradigm of CAR targeting of a shared tumor and tissue differentiation antigen resulting in untenable on-target npg p e r s p e c t i v e nature medicine VOLUME 22 \| NUMBER 1 \| JANUARY 2016 but off-tumor toxicities is GD2. In a series of pediatric patients with GD2expressing neuroblastomas, infusion of a GD2-specific CAR resulted in tumor regression in 3 of 11 patients with active disease, including two sustained CRs 73 . No patients developed neuropathic pain, although several experienced somatic pain at the tumor site. These results are currently being confirmed in two clinical trials (NCT02107963 and NCT01822652) using a new GD2-specific CAR design. Targeting of shared antigens. As previously outlined in detail 74 , a similar pattern of on-target but off-tumor toxicities have been observed with gene-engineered TCRs targeting shared tumor and tissue differentiation antigens. This includes TCRs reactive against MART 75 , gp100 (ref. 75) and carcinoembryonic antigen (CEA) 76 . Collectively, these data suggest that targeting shared tissue differentiation antigens is likely to come with the price of toxicity to critical normal organs. Whereas toxicities related to CAR targeting of hematologic antigens such as CD19 are manageable with repletion therapies, the majority of solid tumors are not derived from 'expendable' or replaceable tissues. One potential solution to minimizing undesired on-target but offtumor toxicities is the engineering of safety and tissue-selectivity mechanisms into the transferred T cells (Box 1 and Fig. 2). A second solution to overcoming the limitation of on-target but off-tumor toxicity is the targeting of antigens uniquely expressed by tumor cells. The ideal antigen would be expressed in common across multiple tumor types and result from driver mutations that appear early in oncogenesis, directly contribute to the malignant phenotype, and are essential for cancer cell survival 77 . Examples could include epitopes encompassing hot-spot mutations in genes such as KRAS, NRAS, ALK, PI3K and BRAF, among many others. Similarly, epitopes derived from virally encoded genes, such as the human papillomavirus (HPV) E6 and E7 oncoproteins which cause cervical, anal, head and neck cancers, might also represent excellent targets because expression of these antigens is exclusive to cancer cells and not normal tissues 74 . An example of a shared mutation in a solid cancer currently being targeted in immunotherapy clinical trials is EGFR variant III (EGFRvIII), a mutated version of EGFR resulting from an in-frame deletion of exons 2 to 7 of the gene 78 . This rearrangement leads to constitutive activation of the cell surface receptor and formation of an immunogenic epitope. Because EGFRvIII is expressed in approximately 30% of patients with glioblastoma multiforme, the most common Box 1 Engineering safety and tissue-selectivity mechanisms into transferred T cells. Complementing the identification of tumor-specific antigen receptors, strategies are being developed that enhance either the safety or tissue-selectivity of engineered T cells (Fig. 2). Suicide genes. T cells can be modified with a 'suicide' gene that confers sensitivity to a prodrug or antibody administered in the case of an adverse event. For example, insertion of herpes simplex virus-thymidine kinase (HSV-TK) renders T cells susceptible to the antiviral medication ganciclovir 135 . HSV-TK is the most extensively tested suicide gene in humans, and it was originally developed to modulate graft-versus-host disease (GvHD) after allogeneic transplantation 135 or to attenuate immunopathology after ACT into immune-deficient hosts 136 . Because HSV-TK is a highly immunogenic virus-derived protein, cells expressing it can be immunologically rejected thereby compromising cellular persistence 136 . Alternative cell-suicide strategies have been developed using constructs derived from human proteins to reduce the risk of immune depletion. In the inducible caspase-9 (iCasp9) system, the sequence of human caspase-9 deleted for its endogenous activation domain is grafted onto a modified variant of human FK506-binding protein 137 . This allows for dimerization and activation of apoptosis upon ligation with a dimerizer drug. The in vivo activity of iCasp9 has been confirmed in patients with acute GvHD after allogeneic donor lymphocyte infusion 138 . Cells engineered with iCasp9 have been detected in patients for over 2 years after transfer 139 , indicating that the construct is not overtly immunogenic. Truncated EGFR (tEGFR) is another suicide gene derived from human sequences that uses the clinically approved EGFR-specific mAb cetuximab to deplete transduced cells 140 . The tEGFR molecule consists of an extracellular portion of human EGFR containing the epitope recognized by cetuximab. In pre-clinical studies, cetuximab caused antibody-dependent cellular cytotoxicity of tEGFR + T cells. Clinical trials incorporating tEGFR are currently enrolling patients. Spatiotemporal control of receptor activation. Strategies that control the duration, location and timing of engineered receptor activity might also enhance safety. RNA electroporation introduces receptors into T cells with self-limited expression, because RNA does not integrate into a cell's genome and is an inherently short-lived molecule in vivo 141 . In the event of toxicity, receptor expression on transferred cells will extinguish spontaneously within several days. Attempts to control the activation of receptor-engineered T cells have also been made. In one design termed logic-gated CARs, T cells are co-transduced with two separate CARs: one that provides suboptimal activation when stimulated alone, and a second that recognizes a separate tumor-associated antigen, which provides a co-stimulatory signal 142 . Thus, only simultaneous ligation of both receptors will allow T cell activation, providing an additional degree of specificity. Alternatively, CARs delivering a dominant antigen-specific inhibitory signal, termed inhibitory CARs (iCARs), have also been explored 143 . The generation of an iCAR is accomplished by attaching the signaling domains of co-inhibitory receptors such as CTLA-4 or PD-1 to a scFv that recognizes structures on normal tissues. In vitro, iCARs suppress cytokine release, cytotoxicity and cellular proliferation after exposure to targets that co-express antigens recognized by both the stimulatory receptor and the iCAR. Recently, the ability to selectively enhance antibody binding within the tumor microenvironment has been demonstrated using pro-antibodies 144 . In this design, the antigen-binding domain of an antibody is sterically blocked by a substrate peptide cleaved in the presence of matrix metalloproteinases enriched within the tumor microenvironment. Provided that the rate of CAR T cell recirculation is minimal once a receptor is unblocked, the use of these 'masked' CARs provides another means of focusing the activity of gene-engineered cells toward targets shared with healthy tissues. Finally, 'on-switch' CARs have been developed 145 . Here, the antigen-binding and intracellular-signaling domains of a receptor are separated into two components that assemble only in the presence of a small-molecule dimerizer. Thus, this system allows for pharmacologic control over CAR T cell activity. It is important to note that none of the spatiotemporal control mechanisms listed above have been tested in humans to date. npg p e r s p e c t i v e and deadly primary adult brain tumor, it could represent an ideal immunotherapy target. Ongoing trials are testing the antitumor activity and safety of 2 different anti-EGFRvIII CAR designs 79,80 ; however, it is currently too early to know whether these receptors will have clinical activity. Similarly, a TCR trial targeting an A2-restricted epitope derived from the E6 oncoprotein of the high risk HPV-16 serotype has also been initiated in patients with HPV-16 + malignancies 81 . Another example of a group of antigens potentially shared across multiple tumors and not present on the surface of normal tissues are the CGAs. As noted above, the CGAs are a group of more than 100 potentially immunogenic proteins encoded by non-mutated genes whose expression in adult tissues is typically restricted to non-MHCbearing germ cells 37 . During the genetic and epigenetic dysregulation leading to oncogenesis, CGA expression may be reactivated causing expression in a variety of cancer types. Most studies evaluating the frequency of CGA expression in cancer have used mRNA-based assays, such as quantitative PCR (qPCR). By using this technique, expression estimates approaching 50% or higher for patients with a given tumor type have been reported for many common epithelial malignancies, including bladder, esophageal, hepatocellular, non-small-cell lung and gastric cancers 37,82 . However, qPCR does not assess the uniformity of CGA expression among individual tumor cells within a patient. When protein expression techniques such as immunohistochemical (IHC) staining are used, strong expression of CGAs such as MAGE and NY-ESO-1 is often limited to only small numbers of cancer cells 37,83 . For example, in the original description of NY-ESO-1 expression by IHC 83 , only 2 of 13 tumors that showed expression as measured by qPCR also demonstrated protein expression in >50% of cancer cells, Administration of a small molecule dimerizer induces dimerization and activation of iCasp9 that subsequently triggers executioner caspases-3, -6 and -7, resulting in apoptosis. (c) Co-expression of a truncated variant of human EGFR (tEGFR) in receptor-modified T cells. Infusion of the EGFR-specific mAb cetuximab results in antibody-dependent cellular cytotoxicity of tEGFR + T cells once the antibody's Fc domain is engaged by Fc gamma receptors (FcγR) on the surface of immune effector cells. Some reports also suggest that cetuximab might deplete tEGFR + cells though complement fixation, but this mechanism of action remains controversial. MAC, membrane attack complex. (d) RNA electroporation of antigen receptors into T cells. Because of the short half-life of RNA species, receptor expression is self-limited after cell transfer, thereby restricting the potential for uncontrolled on-target but off-tumor toxicities. (e) CARs engineered to deliver a dominant antigen-specific inhibitory signal, termed inhibitory CARs (iCARs), work by attaching the signaling domains of co-inhibitory receptors to an antibody-binding region that recognizes structures on the surface of normal tissues. (f) CARs can be 'logic-gated' by co-transducing with two separate CARs: one that provides suboptimal activation when stimulated alone and a second that recognizes a separate antigen, which provides a co-stimulatory signal. Ligation of either receptor alone is insufficient to trigger T-cell activation whereas co-engagement allows T cells to proliferate, acquire effector functions, and exhibit on-target immunity only against tissues expressing both antigens. (g) CARs can be engineered with an 'on-switch', whereby the antigen-binding and cytosolic-signaling domains of the receptor are divided into distinct modules. Administration of a small-molecule dimerizer induces heterodimerization of these modules, initiating cellular activation only when both cognate antigen and the small molecule are present. Thus, the duration and intensity of receptor-engineered T cells can be controlled. (h) CARs can be engineered with a 'masked' receptor in which the antigenbinding domain is sterically blocked by a peptide mask attached to the receptor by a protease-cleavable linker. Entry of modified cells into tissues enriched in proteases, such as the tumor microenvironment, can cleave the blocking peptide and unmask the binding capacity of the CAR. npg a finding confirmed by others 37 . This stands in sharp contrast to what is typically observed with tissue differentiation antigens, in which antigen expression is both more common and more pervasive 84 . From an immunotherapy perspective, treating patients with heterogeneous expression of a target antigen raises concerns about applying selection pressure for antigen-negative tumor cells. Indeed, in all published clinical trials in which objective clinical responses were observed after ACT of cells targeting CGAs, the patients had cancers with a high intensity of antigen expression in the majority (>50%) of cancer cells 36,85,86 . Whether ACT targeting a single antigen expressed on ≤50% of tumor cells can cause cancer regression is the subject of current pre-clinical and clinical investigations. It is important to note that despite their name and the absence of off-target toxicities in trials testing an A2-restricted NY-ESO-1 TCR 36,87 , not all CGAs are exclusively cancer or germline specific. PRAME (preferentially expressed antigen in melanoma), for example, is a CGA with well-documented expression in healthy tissues such as the adrenal glands, placenta and endometrium 88 . Further, whereas the CGA MAGE-A3 appears not to be expressed in healthy tissues, MAGE-A12 and possibly other MAGE family members are expressed at low levels in selected areas of the brain 86 . This became tragically apparent when patients received ACT using T cells modified with an HLA-A2-restricted, affinity-enhanced MAGE-A3 TCR that crossreacted with a non-identical A2-restricted MAGE-A12 epitope 89 . Among nine treated patients, four developed neurologic toxicities, including two neurologic-related deaths 86 . In a separate trial testing the efficacy of an HLA-A1-restricted, affinity-enhanced MAGE-A3specific TCR, two cardiovascular-related deaths occurred 90 . Although MAGE-A3 expression was not detected in the heart, engineered T cells were found to cross-react with an unrelated peptide derived from the muscle-specific protein titin 91 . Further limiting the number of potential patients who might benefit from TCR targeting of CGAs is the requirement that patients possess a HLA haplotype compatible with the TCRs available. Given that the most common MHC class I restriction elements used for TCR gene therapy studies, namely HLA-A1 and A2, are-at most-present in only about 15-50% of patients 92 , the number of patients who would be candidates for these TCR-engineered cells becomes incrementally smaller. Taken together, these data suggest that although TCR-engineered cells targeting certain CGAs can cause pronounced tumor regression without off-target toxicity in the subset of patients whose tumor cells uniformly express the target antigen and who possess a requisite HLA haplotype, the majority of patients with common solid cancers will not be candidates for such therapies. Targeting private somatic mutations with autologous gene therapy Evidence supporting immunogenic neoantigens. After a decadeslong effort by our group and others targeting shared tissue differentiation antigens with cancer vaccines 93 and gene-engineered T cells 74 , cancer immunologists are reassessing which antigens are responsible for immune-mediated cancer regression. Increasing clinical evidence supports the hypothesis that immunogenic products of somatic mutations unique to each patient's cancer-so-called neoantigens-are the relevant targets for successful immunotherapies 13,94 . Neoantigens may represent ideal targets because somatic mutations are central to the formation of most cancers; in other words, mutations may be functionally important to drive tumor growth and/or invasion. Further, neoantigens are exclusive to tumor cells, minimizing the risk of on-target, off-tumor killing of healthy tissue. Finally, because the mutations from which neoantigens are derived are somatic, the repertoire of TCRs expressed on T cell progenitors do not encounter neoantigens during thymic development and therefore should not be deleted by negative selection. Consequently, TCRs with high affinity for neoantigens may be present in the circulation. Several lines of evidence support these suppositions. First, melanoma TILs mediate curative responses with minimal autoimmune sequelae 14 . By contrast, transfer of gene-engineered T cells expressing high-affinity receptors for pigment antigens resulted in suboptimal antitumor responses and severe on-target but off-tumor toxicities 75 . Thus, reactivity to targets besides differentiation antigens appears responsible for a substantial portion of melanoma TIL antitumor efficacy. Consistent with this premise, high-throughput assays using large collections of MHC multimers loaded with shared antigens revealed a low frequency of TILs expressing TCRs that bind to pigment and cancer-germline antigens 95 . By contrast, TILs that release inflammatory cytokines in response to patient-specific neoantigens can be detected at much higher frequencies in the majority of evaluated patients 15,16 . Second, it seems that immune checkpoint inhibitors are particularly effective in cancers with high burdens of somatic mutations. This includes diseases such as melanoma, which carries a high mutation load as a result of UV exposure 10,96 , tobacco-associated lung cancer 11 , bladder cancer 5 and cancers arising in patients with defects in DNA mismatch-repair enzymes 8 . Indeed, in people with melanoma and lung cancer, the mutational burden and neoantigen load is highly correlated with clinical benefit from treatment with blocking antibodies specific to CTLA-4 and PD-1 (refs. 10,11,96). Third, the frequency of circulating neoantigen-specific T cells increased in responding patients following treatment with antibodies specific to CTLA-4 (refs. 9,10) or PD-1 (ref. 11) Finally, we recently demonstrated that a single infusion of a near-clonal population of neoantigen-specific CD4 + T cells resulted in prolonged tumor regression in a patient with cholangiocarcinoma 18 . Taken together, these data suggest that the isolation and re-infusion of neoantigen-specific T cells might be required to mediate tumor regression without inducing undesired on-target but off-tumor toxicities. If the TCRs necessary to induce cancer regression target neoantigens resulting from private somatic mutations unique to each patient's cancer, is individualized gene-engineered T cell therapy possible? We believe that with proper investment and sufficient technologic and regulatory innovation, the answer can be yes. Further, as outlined below it may be possible to assemble all the necessary elements required to generate such an autologous TCR gene therapy solely from the peripheral blood (Fig. 3). Identifying neoantigens and neoantigen-reactive T cells. Currently, elucidation of the mutational landscape of a patient's tumor as a first step in the detection of neoantigen-specific T cells is accomplished by performing whole-exome sequencing (WES) and/or RNA sequencing (RNA-seq) on tumor cells obtained through a tissue biopsy or surgical resection of a metastatic focus 15,16,18,19 . However, it is now possible to perform WES on circulating tumor cells (CTCs) 97 or the cell-free products of tumor cells present in the circulation, such as tumor DNA 98 . These 'liquid biopsies' not only obviate the requirement for a patient to undergo an invasive procedure, they also can detect mutations held in common between the primary tumor and metastatic deposits 97 . Such founder or 'trunk' mutations are likely to be expressed in all tumor cells and therefore would make excellent immunotherapy targets. That said, not all detected mutations will be immunologically recognized as a cancer antigen. For a mutation to be immunogenic, it must be contained in an expressed gene, include a change in amino acid sequence resulting from either a nonsynonymous npg p e r s p e c t i v e substitution or translocation, and be processed intracellularly into a 9-to 11-amino acid peptide capable of binding to one of the patient's MHC molecules. Therefore, high-throughput methods for screening and isolating neoantigen-reactive T cells after a patient's tumor has been sequenced are required. In one approach, synthetic long peptides that incorporate the substituted amino acid from a nonsynonymous mutation and that are predicted to be strong binders to one or more of the patient's MHC molecules are pulsed onto autologous antigen-presenting cells (APCs) 15,16 . In a second approach, a minigene encoding the mutated amino acid and flanked on either side by 12 amino acids from the endogenous protein can be electroporated into autologous APCs [17][18][19] . With either technique, establishment of a patient-specific tumor cell line is not required. One can use different sources of T cells to screen for neoantigenreactive T cells. If a patient is able to undergo a metastatic resection, TILs can be isolated from this specimen, expanded and used as the screening population. Neoantigen-reactive TILs upregulate expression of tumor necrosis factor (TNF) family co-stimulatory markers such as 4-1BB and OX- 40 (refs. 15,18,19,99,100), and/or the degranulation marker CD107a (ref. 100), when co-cultured with antigen-bearing APCs or autologous tumor cell lines. Such receptor upregulation can be used to identify neoantigen-specific T cells and to isolate these cells (either by fluorescence-activated cell sorting (FACS) or magnetic bead isolation) for further analysis 100,101 . Similarly, although PD-1 is a negative regulator of T cell functions, expression of this molecule also marks TILs that react with neoantigens 99 . Therefore, isolation of PD-1 + TILs can also be used to enrich for neoantigen-reactive T cells. Detection of neoantigen-specific TILs using these techniques can be successful not only in cancers with high mutational loads, such as melanoma, but also in cancers harboring relatively few mutations. For example, we recently reported that neoantigen-specific T cells could be identified from the TILs of nine out of ten sequentially screened patients with various gastrointestinal malignancies 19 . None of these patients had mutations in DNA mismatch-repair enzymes nor did they possess a microsatelliteunstable phenotype associated with a high mutational burden. Critically, a neoantigen-specific T cell population could be identified Figure 3 A pathway for generating autologous TCR gene therapies targeting neoantigens for patients with advanced epithelial cancers. From a single blood draw, all of the requisite components required to produce this therapy can be procured. Circulating tumor cells (CTCs) can be enriched from the blood using a combination of antibody-mediated isolation based on epithelial marker expression (for example, EpCAM) followed by microfluidic isolation. Subsequent genomic extraction, amplification and whole exome sequencing can identify non-synonymous mutations present within the tumor. Circulating T cells that express PD-1 can be isolated and co-cultured with autologous professional antigen presenting cells (APCs) that have either been pulsed with synthetic long peptides harboring the amino acid change resulting from the mutation or RNA-electroporated with tandem minigenes (TMGs) encoding the amino acid change and flanked on either side by 12 amino acids from the endogenous protein. T cells that upregulate the activation markers 4-1BB and/or OX-40 can then be isolated and the α and β chains of the TCR associated with this cell can be sequenced. The α/β TCR that confers reactivity against a neoantigen can then be cloned into an expression vector, for example an integrating virus or the Sleeping Beauty (SB) transposon/transposase system. T cell subsets isolated from the peripheral blood of the patient can finally be modified with this expression vector, expanded in vitro to numbers sufficient for treatment, and re-infused back into the patient. TL, transmitted light; T N , T naive; T SCM , T stem cell memory; T CM , T central memory; SB11, SB transposase 11. Select reactive T cells by 4-1BB and/or OX-40 upregulation APC npg in a patient with pancreatic cancer whose tumor harbored only ten somatic mutations. These data demonstrate that it is possible to consistently identify neoantigen-specific TIL cells in common epithelial cancers that are not hyper-mutated and that seem to be unresponsive to other forms of immunotherapy, such as checkpoint inhibitors 8 . In addition to TILs, neoantigen-specific T cells can also be identified directly from the peripheral blood of a cancer patient. This can be accomplished using a high-throughput methodology based on peptide-MHC complexes 9,11,102 . In this approach, proteasomal processing and HLA-binding algorithms are used to determine candidate epitopes containing nonsynonymous mutations predicted to bind with high affinity to one of the patient's MHC alleles. Predicted peptides are subsequently synthesized and peptide-MHC complexes can be generated by UV-induced peptide exchange reaction 103 . By using this technique, it was shown in a recent report that it was possible to reproducibly identify, isolate and expand neoantigen-specific T cells from starting frequencies as low as 0.002% (ref. 102). Complementing this approach, it is probable that PD-1 may also identify tumorreactive T cells in the circulation, just as it enriches for mutationspecific T cells in TILs 99 . Engineering neoantigen-reactive T cells. The mere isolation of mutation-specific and tumor-reactive T cells does not necessarily lead to an effective cancer treatment. This is because expansion of small numbers of isolated T cells to therapeutic numbers can result in a loss of replicative capacity and entry into a terminally differentiated state 104 . The development of effective T cell therapies from one or a limited number of cells might be possible with cellular reprogramming techniques 105 . The use of induced pluripotent stem cell (iPSC) technology with T cells has been experimentally accomplished 106,107 . Importantly, reprogrammed T cells retain a rearranged antigen-specific TCR. Although early reports suggested that iPSC-derived cells may be at increased risk for immunologic rejection after transfer into syngeneic hosts 108 , these findings have not been universally observed 109,110 . Other technical and practical hurdles remain in realizing the clinical utility of such T cell reprogramming techniques, however. For example, the efficiency of reprogramming remains low 111 , reprogrammed T cells can have an innate-like CD8αα or γδ T cell phenotype 107,112 , and differentiated cells derived from induced pluripotent cells might be at increased risk of undergoing malignant transformation 113 . Addressing each of these issues remain brisk areas of investigation. An alternative approach using currently available technologies to generate minimally differentiated antitumor T cells could be cloning mutation-specific TCRs followed by inserting these TCRs into selected T cell subsets. This can be accomplished using high-throughput TCR gene-capture 100 or multiplex nested single-cell real-time PCR (RT-PCR) 114 . More recently, it has been possible to perform α/β paired TCR sequencing (pairSEQ) from bulk populations of T cells using combinatorics 115 . After a mutation-specific α/β TCR has been identified, it subsequently can be cloned into a good manufacturing process (GMP)-quality expression system. In this case, substantial innovations in regulatory oversight are required. It presently costs as much as $250,000 and a takes a minimum of 4-6 months to generate a GMP retroviral or lentiviral vector 116 . Both this expense and time scale are prohibitive for treating individual patients, although it might be feasible to simplify safety testing for human gene therapy trials 117 . A major contributor to the time and expense in generating a GMP viral vector is mandated biosafety testing. Whereas testing for sterility and the presence of mycoplasma is rapid and relatively inexpensive, testing for adventitious viruses, species-specific viruses, and replication-competent retrovirus (RCR) is not. Indeed, the necessity for routine RCR testing has been questioned by a large group of academic investigators involved in clinical gene therapy trials 117 . Given the potential for a favorable benefit/risk ratio in this patient demographic if autologous gene therapies are highly effective, it might be ethically permissible to relax some routine biosafety testing requirements. Ultimately, resolution of these issues will require close collaboration among all stakeholders, including regulatory agencies. As an alternative to retroviral and lentiviral vectors, nonviral genetransfer methods can potentially be used to genetically introduce antigen receptors. Because nonviral integration systems use oligonucleotides and recombinant proteins, they can be considerably cheaper to manufacture and easier to implement for single-use applications compared with viral vectors. By some estimates, production of nonviral reagents may cost one-tenth that of GMP-grade virus 118 . Presently, use of the Sleeping Beauty (SB) transposon/transposase system has advanced farthest in clinical development 22 . Genome editing strategies that introduce double-stranded DNA breaks that serve as sites for targeted gene insertion, including zinc-finger nucleases, transcription activator-like effector nucleases (TALENs), and the clustered regularly interspersed short palindromic repeat (CRISPR)-Cas9 system, might offer additional nonviral means of inserting antigen receptors in the near future 119 . Although the SB transposon/transposase system typically has a lower gene-transfer efficiency compared with viral integration techniques, selection for modified cells by drug exposure 120 , magnetic bead isolation 121 , or propagation using artificial antigen presenting cells 122 can improve the frequency of receptor-engineered cells. Finally, it will probably be advisable to enrich for 'younger' T cell subsets with high proliferative and engraftment potentials, such as naive, stem cell memory and central memory T cells, before introduction of the receptor 104,123 . In preclinical studies, these subsets are superior to more-differentiated effector memory and effector T cell subsets in mediating cancer regression 124 . Isolation of defined T-cell populations can be accomplished using antibody-microbead conjugates 125 or streptamer 126 isolation strategies. Each of these activities, including cell processing, genetic engineering and cell expansion, can be conducted at one of several centralized GMP cell manufacturing facilities within the US. These include facilities that were previously used to generate the sipuleucel-T product 23 , as well as newer facilities that are generating CD19-specific CARs. Limitations targeting neoantigens. Even with resolution of the technical and regulatory barriers listed above, critical questions regarding the targeting of neoantigens using patient-specific receptors remain. First, it is unclear how many receptors targeting distinct antigens might be required to reliably induce responses. Clinical experience using ACT of CD19-specific CAR-and NY-ESO-1-specific TCR-modified cells demonstrates that it is possible to induce durable CRs by targeting a single epitope 36,38,41,47 . However, this experience might not be transferable to the targeting of antigens generated by somatic mutations. Heterogeneity of the mutational landscape within a tumor mass and between metastases has been well documented in solid cancers 127,128 , raising the possibility that not all cancer cells within a single tumor or within a single patient will express the cognate target for a neoantigenspecific receptor. Under the selective pressure of an antigen-specific immune response, outgrowth of cancer cells lacking the target antigen might occur, a phenomenon recently documented in treatment failures with CD19-specific CAR cells 129 . Preclinical ACT studies in mice give conflicting conclusions as to whether direct cell killing of npg p e r s p e c t i v e individual cancer cells expressing a target antigen is required to induce tumor regression, or whether bystander destruction is sufficient [130][131][132] . For example, epitope spreading has been proposed as a mechanism by which T cell killing of a limited population of tumor cells can lead to immunologic destruction of other tumor cells expressing unrelated antigens. There are several reported examples of patients receiving ACT in which transfer of a mono-specific T cell population resulted in increased immune reactivity against unrelated tumor or viral antigens 85,133 . Evidence for epitope spreading has also been documented in some patients responding to vaccine immunotherapies 134 , however it is not known whether this phenomenon is causally associated with tumor regression. It would therefore seem wise to target multiple antigens using a panel of different receptors. Similarly, targeting neoantigens derived from the products of a driver or truncal mutation expressed in the primary tumor might reduce the risk of targeting a subpopulation of clonally divergent tumor cells. Concluding remarks Genetic redirection of patient T cells toward the B cell lineage antigen CD19 using CD19-specific CARs has met with unprecedented success in a heterogeneous group of chemotherapy-refractory hematologic diseases. So successful is this approach that it is very likely to become part of the standard of care for these diseases in the near future. Although this approach comes with the toxicity of normal B cell depletion, most patients can survive with Ig repletion. Nevertheless, the simple extrapolation of the experience using CARs in the treatment of lymphoid cancers to the treatment of most solid tumors is likely to come with the steep price of toxicity to essential normal tissues. Success for cell-based immunotherapies may come from the arduous task of targeting the unique set of mutations that cause each patient's cancer.	2017-11-08T19:13:48.663Z	2016-01-01T00:00:00.000	{ "year": 2016, "sha1": "465942718241be05b78ab5e158877d294b182027", "oa_license": "CC0", "oa_url": "https://zenodo.org/record/1233455/files/article.pdf", "oa_status": "GREEN", "pdf_src": "SpringerNature", "pdf_hash": "368368f6109243b8a4d74d64f9d7d6e06d473220", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
18800268	pes2o/s2orc	v3-fos-license	Asymptotics for the Eigenvalues of the Harmonic Oscillator with a Quasi-Periodic Perturbation We consider operators of the form H+V where H is the one-dimensional harmonic oscillator and V is a zero-order pseudo-differential operator which is quasi-periodic in an appropriate sense (one can take V to be multiplication by a periodic function for example). It is shown that the eigenvalues of H+V have asymptotics of the form \lambda_n(H+V)=\lambda_n(H)+W(\sqrt n)n^{-1/4}+O(n^{-1/2}\ln(n)) as n\to+\infty, where W is a quasi-periodic function which can be defined explicitly in terms of V. Introduction The one-dimensional harmonic oscillator is the operator where α is a positive parameter. We can consider H as an unbounded self-adjoint operator acting on L 2 (R). The determination of the spectrum of H is a classical problem -virtually any introductory book on quantum mechanics has a section devoted to this topic. In particular H has a compact resolvent and hence a discrete spectrum. Furthermore, the eigenvalues of H are simple and can be enumerated as λ n (H) = α(2n + 1), n ∈ N 0 . The purpose of this paper is to study the large n asymptotics of the eigenvalues of the perturbed operator H + V when V is a self-adjoint quasi-periodic pseudodifferential operator of order 0. More precisely, we assume V can be written in the form where Λ ⊂ T * R ∼ = R 2 is a countable discrete index set and, for each a = (a x , a ξ ) ∈ T * R, we define U a to be the unitary operator on L 2 (R) given by U a φ(x) = e iaxa ξ /2 e iaxx φ(x + a ξ ). The V a 's are just complex coefficients. Since U * a = U −a for any a ∈ T * R, the condition that V is self-adjoint can be rewritten as the requirement a ∈ Λ =⇒ −a ∈ Λ and V −a = V a , a ∈ Λ. We will also assume the V a 's satisfy the following condition (essentially a regularity assumption); a∈Λ \|a\| 3 \|V a \| < +∞. (4) In particular, this condition ensures that the right hand side of (2) is absolutely convergent in operator norm, making V a well defined bounded operator. Since H has a compact resolvent the same must then be true for H + V ; it follows that the spectrum of H + V also consists of discrete eigenvalues. Remark. If we take Λ = {(ωm, 0) \| m ∈ Z} then V is the operator of multiplication by a function with period ω whose m-th Fourier coefficient is simply ω 1/2 V (ωm,0) . Condition (4) becomes a standard regularity requirement (that the function V should be a "bit more" than C 3 ). In general we may consider V to be a zero-order pseudo-differential operator with Weyl-symbol a∈Λ V a e i(axx+a ξ ξ) (n.b., U a is the operator with Weyl-symbol e i(axx+a ξ ξ) ). If Λ is a rational periodic lattice then V will be a periodic operator (in the sense that it commutes with a specific translation operator). Taking Λ to be an irrational periodic lattice, or an irregular discrete set, leads to a generalisation of such periodic operators; when we apply "quasi-periodic" to V we mean this particular type of generalisation. If 0 ∈ Λ then the corresponding term in V is V 0 times the identity operator and will thus cause a simple shift in the spectrum of H by V 0 . This term is included in the statement of the main result (Theorem 1.1 below) but thereafter we shall assume V 0 = 0. We also set Λ ′ = Λ \ {0}; since Λ is discrete, T * R \ Λ ′ contains a neighbourhood of 0. Define a metric \|·\| α on T * R by \|a\| α = (α −1 a 2 x + α a 2 ξ ) 1/2 . This metric is equivalent to the usual metric \|·\| so condition (4) can be rewritten as The main result of the paper is the following. Theorem 1.1. Suppose V given by (2) satisfies (4) (or equivalently (5)). Then the eigenvalues of the operator H + V satisfy as n → ∞, where W : R → R is the quasi-periodic function defined by The presence of the quasi-periodic function W means the first order asymptotics given by Theorem 1.1 contain considerably more information about the operator V than one might expect (c.f. the simple power type asymptotics for the case when V is given as multiplication by an element of C ∞ 0 ( [PS]) or for the operator −d 2 /dθ 2 +V (θ) on S 1 (see Theorem 4.2 in [MO])). In particular we note that if V is given as multiplication by a periodic function, knowledge of the first order asymptotics of λ n (H + V ) allows the Fourier coefficients of V to be "half" determined (the values of V (−mω,0) + V (mω,0) , m ∈ N, can be determined from W ). It is likely that there exists a full asymptotic expansion for λ n (H + V ), involving further terms with quasi-periodic functions multiplying increasingly negative powers of n. Judging by numerical evidence (for example with the potential V (x) = cos(x)) the second term in the asymptotics is O(n −3/4 ). This order (even as an improvement of the remainder estimate in Theorem 1.1) appears to involve reasonable subtle cancellation effects within the series giving the second term of the asymptotics; no attempt to deal with this analysis is made here. Remark. With an obvious modification to the definition of W and a remainder estimate of O(n −1/3 ln(n)), Theorem 1.1 also holds for operators V of the form In this case V is a pseudo-differential operator of order zero whose Weyl-symbol has Fourier transform 2πV a . The \|a\| 3 α term in the condition on V a is then a regularity condition, while the \|a\| −3/2 α term is a generalisation of quasi-periodicity. The proof of Theorem 1.1 is given in Section 4 using standard ideas to express the eigenvalues of H + V in terms of a series involving the resolvent of H and the operator V . The non-triviality of Theorem 1.1 is contained in technical results used to establish the convergence of these series. These results are obtained in Sections 2 and 3; estimates for the elements V φ k , φ k ′ of the matrix of V with respect to the eigenbasis {φ k \| k ∈ N 0 } are obtained in the former and are then combined to give resolvent estimates in the latter. Notation. We use C to denote any positive real constant whose exact value is not important but which may depend only on the things it is allowed to in a given problem. Appropriate function type notation is used in places to make this clearer whilst subscripts are added if we need to keep track of the value of a particular constant (e.g. C 1 (V ) etc.). We use T , T 1 and T 2 to denote the operator, trace class and Hilbert-Schmidt norms of the operator T respectively. Estimates for Matrix Elements The aim of this section is to obtain the necessary estimates for the matrix elements V φ k , φ k ′ for all k, k ′ ∈ N 0 . In turn these will be estimated via defined for all a ∈ T * R and k, k ′ ∈ N 0 . Since the operator U a is unitary we immediately get \|U k,k ′ a \| ≤ 1. To obtain more precise estimates we can use the following special function identity (see 7.377 on page 844 of [GRJ]) to find an explicit formula for U k,k ′ a ; for any 0 ≤ k ≤ k ′ and y, z ∈ C we have where L Proof. Introduce the complex number From (7), (3) and (1) we get where the last line follows from (9). Now \|ω\| = ρ while The result follows. Throughout the remainder of this section we will assume a ∈ T * R\{0} is fixed and ρ > 0 is given by (10). Laguerre polynomials can be expressed in terms of the confluent hypergeometric function; using 22.5.54 in [AS] we get The confluent hypergeometric function can, in turn, be written as a pointwise absolutely convergent series of Bessel functions; from 13.3.7 in [AS] we get and, for j ≥ 2, It follows from Lemma 2.1 that The next two results give estimates for the constants appearing in (13). where the numerator and denominator both contain If k ′ − k is odd this can be rearranged as The result now follows from the fact that for any 0 ≤ m ′ ≤ n. Next we obtain some estimates for the Bessel functions appearing in (13). Now combining where the last line follows from the hypothesis that ρ(k ′ + k + 1) −1/6 ≤ 1/2. Lemma 2.5 can now be used to estimate the remaining Bessel function term. First order term The next result is used to obtain the explicit form for the first order correction term in the asymptotics for λ n (H + V ). Resolvent Estimates For any λ ∈ C\σ(H) let R(λ) = (H − λ) −1 denote the resolvent of the operator H; we will also write R for R(λ) where this should not cause confusion. The first two results in this section relate to the operator R(λ)V R(λ), which is clearly bounded whenever λ is in the resolvent set of H. We show that it is in fact trace class while its operator norm decreases as n −1/4 for λ ∈ Γ ε,n . Lemma 3.1. For any n ∈ N and λ ∈ Γ ε,n we have We remark that since {φ k \| k ∈ N 0 } is an orthonormal basis of L 2 (R) Proof. Using the orthonormal basis {φ k \| k ∈ N 0 } we have We will split this sum using the partition (23). Firstly Proposition 2.7 and (25) imply Now using (27), (28) and (25) we get The remaining part of the sum on the right hand side of (29) involves k ∈ J and k ′ ∈ I ⊂ N 0 ; thus we can estimate this part using an argument similar to the last one with k and k ′ swapped. Proof. The set {φ k \| k ∈ N 0 } is an orthonormal eigenbasis for R with corresponding eigenvalues (λ k − λ) −1 , k ∈ N 0 so (25) implies Suppose n ∈ N 0 and j ∈ N. From the previous result we know that R(λ)V R(λ) is trace class for any λ ∈ Γ ε,n . On the other hand R(λ)V is bounded (in fact R(λ)V ≤ ε −1 V ). It follows that is also trace class with trace norm uniformly bounded for λ ∈ Γ ε,n . The work in the remainder of this section leads to Proposition 3.5 where we obtain an estimate for the trace of an integral of such operators. Taking f (k) = V φ n , φ k we can use (28) and Proposition 2.7 to check that (30) is satisfied. The next result then follows from Lemma 3.3 by use of induction; we can take K = max{ V , C(V ), K ′ } where C(V ) and K ′ are the constants coming from Proposition 2.7 and Lemma 3.3 respectively. Lemma 3.4. Suppose n ≥ 2 and j ∈ N 0 . Then there exists a constant K = K(V, ε) such that for all λ ∈ Γ ε,n we have Proposition 3.5. Suppose n ≥ 2 and j ∈ N. Then where K is the constant from Lemma 3.4. Continuing by induction we get Together with the fact that Rφ k j , φ k 0 = δ k j ,k 0 (λ k 0 − λ) −1 we now get where A(λ) is the meromorphic function Since for any k 0 , . . . , k j−1 ∈ N 0 , we can rewrite the above equation as Integrating around the contour Γ ε,n it follows that The poles of the meromorphic function A(λ) occur at the points λ = λ k for k ∈ N 0 . Since the only such point enclosed by the contour Γ ε,n is λ = λ n , it follows that the only terms in the series (31) which contribute to the right hand side of (32) are those with at least one of k 0 , . . . , k j−1 equal to n. With the help of symmetry we then obtain the identity For any λ ∈ Γ ε,n we have \|λ n − λ\| = ε while by Lemma 3.4. Since the length of Γ ε,n is 2πε we finally get completing the result. Taking j = 1 in (33) leads to the formula This is needed to obtain the first order correction term in Theorem 1.1. The argument can be tied together using a standard resolvent expansion. Set R V (λ) = (H + V − λ) −1 and let n ≥ N. Then The right hand side of (35) will still converge if V is replaced with gV for some g ∈ [0, 1]. Hence σ(H + gV ) ∩ Γ ε,n = ∅. Since the eigenvalues of H + gV depend continuously on g, it follows that Γ ε,n must enclose λ n (H + V ) but no other points of σ(H + V ). Thus we can write R(λ)(−V R(λ)) j dλ = Tr T n = V φ n , φ n + O(n −1/2 ln(n)). Theorem 1.1 now follows from Proposition 2.8.	2014-10-01T00:00:00.000Z	2003-12-04T00:00:00.000	{ "year": 2003, "sha1": "aa8d4b8583847f0bd9595166b76a60d5a8d10386", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "78939f2f69b9311ae467abeb17bee276b5b46436", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics" ] }
247545088	pes2o/s2orc	v3-fos-license	“We threw away the stones”: a mixed method evaluation of a simple cookstove intervention in Malawi [version 2; peer review: 2 approved] Background: Air pollution exposure is responsible for a substantial burden of respiratory disease globally. Household air pollution from cooking using biomass is a major contributor to overall exposure in rural low-income settings. Previous research in Malawi has revealed how precarity and food insecurity shape individuals’ daily experiences, contributing to perceptions of health. Aiming to avoid a mismatch between research intervention and local context, we introduced a simple cookstove intervention in rural Malawi, analysing change in fine particulate matter (PM 2.5 ) exposures, and community perceptions. Methods: Following a period of baseline ethnographic research, we distributed ‘chitetezo mbaula’ , locally-made cookstoves, to all households (n=300) in a rural Malawian village. Evaluation incorporated village-wide participant observation and concurrent exposure monitoring using portable PM 2.5 monitors at baseline and follow-up (three months post-intervention). Qualitative data were thematically analysed. Quantitative analysis of exposure data included pre-post intervention comparisons, with datapoints divided into periods of combustion activity (almost exclusively cooking) and non-combustion periods. Findings were integrated at the interpretation stage, using a convergent design mode well-liked and widely used by residents as substitutes for previous cooking methods (mainly three-stone fires). Commonly cited benefits related to fuel saving and shorter cooking times. Conclusions: The cookstove intervention had no impact on cooking-related PM 2.5 exposures. A significant reduction in background exposures may relate to reduced smouldering emissions. Uptake and continued use of the stoves was high amongst community members, who preferred using the stoves to cooking over open fires. Introduction Air pollution -and fine particulate matter (PM 2.5 ) in particular -is a widely recognised risk factor for cardiorespiratory and wider systemic disease, and the interactions between airborne particulates and climate change also have repercussions for health [1][2][3] . In Malawi, which is largely rural, air pollution is a persistent problem, stemming mainly from domestic cooking: Malawian households cook on average three times per day, using biomass fuel (usually firewood) on three stone fires 4 . Recent ethnographic work on 'smoke' in the Malawian setting highlighted the ways in which local experiences and values -often very different from those of western researchers -can shape locally-relevant priorities for intervention, and contextualised approaches 4 . By centring local perspectives, we can make interventions more context-appropriate, which often also brings benefits in terms of long-term sustainability. For health research which ostensibly aims to improve the lives of people in LMICs, prioritising participants' perspectives -rather than those of researchers -is also arguably best practice 5-7 . In rural Malawi, where experiences of precarity, scarcity and food insecurity are common, these contextual realities often take precedence over externally proposed agendas such as ours. In a recent study exploring Malawian communities' perceptions of health within a trial of advanced cookstoves 8 , participants linked good health primarily to food security 9 . Thus the research imperative in such contexts should be for cleaner air solutions which avoid amplifying existing daily challenges for residents, as well as appropriately addressing shared concerns. In considering options for cleaner cooking in LMICs such as Malawi, economic affordability for the majority is a key consideration [10][11][12][13] . Whilst initial costs of clean stoves are important here, also relevant are costs of ongoing fuel purchase, and maintenance and repair costs of any newly introduced technologies 14-17 . Perceptions of the benefits of new technologies are also context specific. Studies set in various LMIC settings have cited flexibility, in terms of fuel use or place of cooking 18,19 , and ability to cook quickly or for large numbers of people 12,20,21 as important considerations. Whilst cleaner burning biomass-fuelled cookstoves have been largely rejected by health researchers due to suboptimal emission reductions, features such as more efficient fuel use are themselves highly valued by local populations, with consequent potential environmental impacts conferring additional advantage 22 . Thus, while individual household interventions will not be sufficient to achieve clinically impactful reductions in PM 2.5 23,24 there may be wider benefits to adoption of locally relevant cleaner stove types in low-income settings such as Malawi. This could represent a useful interim step on the way to the much-needed provision of clean fuels at scale 25 . Following an extended period of ethnographic and monitoring groundwork in a village in Malawi 4 , we provided locally made clay wood-burning stoves to every household. Realist evaluation aimed to assess residents' views of the cookstoves as well as any changes in personal PM 2.5 exposures three months after cookstove distribution. Ethical considerations The study was approved and sponsored by the LSTM Research Ethics Committee (20-022). In-country ethical approval was granted by the College of Medicine Research Ethics Committee (COMREC) in Blantyre (P.06/20/3069). Informed consent processed were completed for all participants involved in air quality monitoring. For other village residents, an extended process of community consent and introduction was undertaken, with engagement throughout the project ensuring continued consent for participation 1 . Study setting and population The study was set in a rural village of approximately 300 households in Southern Malawi: the site of previous ethnographic and baseline monitoring work 4 . Residents were all subsistence farmers, and economic insecurity was common. Most income came from ad hoc piece work or self-employment Amendments from Version 1 Overall, the paper has been reviewed and changes made in line with reviewer comments, to improve the focus on the qualitative ethnographic findings, and prevent any overstatement of the significance of the quantitative (air quality monitoring) results. More specific changes have also been made in response to reviewer comments. Changes are outlined below: In the abstract and throughout the paper, use of the word 'baseline' has been reviewed. To eliminate confusion, the word is now only used to indicate the pre-intervention phase. In referring to non-combustion exposures, the word 'background' exposures now used for clarity. In the introduction, the word 'persisting' has been replaced with 'persistent', as suggested. Also in the introduction, the following sentence has been rewritten in line with reviewer comments: 'Centring local perspectives in this way, as well as constituting arguably the 'right' approach to global health problems, can optimise the suitability and sustainability of any subsequent solutions' Text has been added under the 'Study setting and population' section to provide more information on the setting and environment, including cooking practices in the village, as suggested by the reviewer. Extra text in the 'Study design and intervention' section and 'Data collection' strengthens this aim. Also under 'Study design and intervention' section and 'Data collection' subheadings, extra text clarifies timescales of the study, which were previously confused. Boxplots (FIgure 4, Figure 5a & Figure 5b) have been updated to plots using corrected values to improve the quality of the figures. Relevant analytic methods have been updated under 'Data analysis'. Quantitative results have been updated slightly to reflect the more accurate re-analysis of the data. Qualitative data have been added to the results in response to reviewer's comments, and elements of the discussion revised in keeping with this. Any further responses from the reviewers can be found at the end of the article REVISED in small businesses. Cooking, mainly carried out by female household members, constituted the main source of PM 2.5 exposure in this setting 26 . Across the village, most cooking was done on three-stone fires, using collected firewood for fuel. Households frequently owned a charcoal cookstove but, as their use required the purchase of charcoal, these were only used on specific occasions, such as when heavy rain prevented the use of three stone fires 4 . In addition, a few houses in the village -two, to our knowledge -had donated firewood cookstoves (or chitetezo mbaula, meaning 'protecting stove'). Residents of these households used the stoves as well as three stone fires, and residents' views on their benefits were mixed. Further contextual details are as previously reported 4 . All households in the village were involved in the participant observation work and the intervention, and in qualitative elements of the evaluation. For exposure monitoring, consenting adult participants were recruited with an aim of achieving a broadly representative sample of village residents, including both men and women, members of different household sizes and structures, and varied cooking needs. These participants had to be resident in the village and habitually spending six or more days per week in the village setting. Children (aged under 18) were not included. Study design and intervention This was a before-after study. Following a period of extended participant observation around the village and individual baseline exposure monitoring in a total of 23 residents (between February and March 2020), all households in the village were given a locally produced firewood cookstove. These moulded, natural-draught cookstoves made of clay were the same as to those already present in a few households, provided by government or non-governmental organization initiatives 27 , and recently piloted in rural Malawi in advance of a large cookstove trial 28 . The cookstoves were introduced to key local representatives (including the chief and a local health surveillance assistant) at a small village meeting, with explanations of their use and some expected benefits, before distribution -without cost -to households, in December 2020. Three months after their initial introduction, researchers (PhD research candidate, SS, and research assistant, HS) returned to the village and continued participant observations around the village, extending between March and May 2021. The originally sampled 23 residents were approached again for involvement in repeat PM 2.5 exposure monitoring (taking place March-April 2021) during the same evaluation period. These methods are depicted in Figure 1 below. Data collection Quantitative data collection. The original sample of 23 participants who took part in air pollution exposure monitoring were asked to each spend a further period of 48 hours carrying personal air quality monitors to assess post-intervention PM 2.5 exposures. PurpleAir PA-II-SD laser particle counting devices (Purple Air, UT, USA) were used, as in the pre-intervention phase, again with 20Ah portable power banks (Anker Innovations, Changsha, China), and carried in specially designed waist bags. The devices took PM 2.5 readings at two-minute intervals throughout the monitoring period. As in the baseline study 26 , on monitor collection, memory cards were removed and the data used to create simple line graphs on a laptop, which were then viewed together, by the participant and researcher, and used as a basis for activity recall. This technique (developed on the basis of earlier work using monitoring alongside participant observations), allowed for division of all traces into 'background' periods of no identified exposure, and periods of 'activity' (where a specific source of combustion was identified). Further information was gathered around each identified episode of cooking, including bathwater warming or fire/stove use for heating, place of cooking, stove or device, and fuel used 29 . Qualitative data collection. Participant observations were carried out by the doctoral researcher (SS) and Malawian research assistant (HS), together with a local fieldworker: a village resident, and centred around cooking activity. As researchers and village residents were familiar with each other, following the initial period of ethnographic participant observation, observations were now spread around the village without the prior focus on a small number of individual households. Researchers visited the village on most days each week over a period of 10 weeks (during the same time period as the second set of exposure monitoring), spending time in all areas of the village over this observation period. Participant observation at this stage involved less active involvement by researchers in daily activities and more passive observation and discussion. Observations were mainly focused around evidence of stoves, fires, food, and fuel use. Discussions, particularly in the post-intervention period, were often based around cooking and related activities (also including food preparation, starting of the fire or cookstove, and washing of dishes), partly because families were most often engaged in these activities when spending time around the household. Discussions were in reality more unstructured, participant-led conversations, and mainly concerned cooking and stove use, although other related topics were incorporated as was felt relevant by participants and researchers. Ad hoc conversations were held with any willing community members who were present at the time of our visits (although care and attention was always given to ethical issues including questions of confidentiality). In view of the social nature of the village setting, these conversations at times involved several women: either from an extended family group, or a group of village residents. At other times conversations were held with individual men and women. Conversations usually took place at residents' homes, almost always outside houses, in yards or verandas. Contemporaneous field notes were made during this fieldwork, integrating discussion content and observations. The study was designed such that pre-and post-intervention monitoring took place during similar months over successive years. Both exposure monitoring periods and the period of post-intervention observation fell during the rainy season in Malawi (which is between between November and April each year). Data analysis Analysis of PM 2.5 exposure data. Descriptive comparisons of proportion of recorded time (datapoints) spent cooking, and specific cooking features (place, device and fuel used) before and after stove introduction were produced. Exposures before and after introduction of the stoves were compared using median and interquartile range values. All exposure datapoints were first divided into 'activity' or 'background', categories using matched time-activity data, and medians and interquartile ranges before and after intervention introduction were then compared for both 'background' and 'activity' subcategories. For boxplots, corrected PM 2.5 values were used: values were log transformed after adding 0.1 to allow log transformation of zero values. For statistical comparisons of pre-and post-intervention exposures, median exposures for each participant (pre-vs post-intervention) were compared using a Wilcoxon signed-rank test. A non-parametric test was chosen as the data did not consistently show a normal distribution 30 . Data were analysed using R 31 , and the package ggplot2 32 was used to create plots. Analysis of participant observation data. Fieldnotes were jointly reviewed and reflected on by SS and HS with input from the local fieldworker, and tentative themes iteratively developed through these discussions. Content of the notes was entered onto QSR NVivo V.12 (released in March 2020) for formal coding (SS) and review (HS). The combination of participant observations with personal monitoring allows a number of benefits including triangulation -avoiding a reliance on 'self-report' by participants -and introducing insights into how interventions work within social contexts 33 : particularly important in the case an intervention centred so firmly in the domestic sphere. The combination of qualitative and quantitative enquiry, with each applied as appropriate, was used here as it allows for a fuller exploration of outcomes, particularly important for complex interventions with social elements 34,35 . Rather than separate but parallel applications and analysis, an integrated synthesis was used, allowing for more in-depth findings than when either single methodology is used alone. Qualitative and quantitative data collection were undertaken concurrently by the same research team, with integration happening at the interpretation stage: the so-called 'Convergent Design' model of mixed method study design 36 . Results post-intervention Between February 2020 and April 2021, 18 participants (15 female; mean age 43, standard deviation 14.2) completed the study with matching pre-and post-intervention traces (February -March 2020, and March -April 2021 respectively). The predominance of women in the sample reflected the majority female nature of cooking in the village. Three participants were lost from the full pre-intervention monitoring set (originally 23 participants) due to participants moving away from village (N=2) and participant death (N=1), and problems with monitors and batteries left only 18 with matching traces. The overall pre-and post-intervention dataset incorporated 1563 hours monitoring time (of which 788 hours post-intervention). In the pre-intervention dataset, trace lengths ranged from 23.3 to 58.5 hours (median 43.1; IQR 39.3 -49.2). Post-intervention traces ranged between 24.1 and 53.9 hours (median 48.6; IQR 40.7 -49.1). Traces shorter than 48H were due to battery faults. Of the total recorded period (pre-and post-intervention), 351 hours (22.5%) constituted 'activity', of which 92% was cooking (including bathwater warming) activity. Other non-cooking activities included exposure to others' fires or stoves (such as when socialising at a neighbour's household) and burning grass on farmland. A larger proportion of the total post-intervention monitoring period constituted combustion activity compared with pre-intervention (30% post-vs. 23% pre-intervention). Further details are available on Harvard Dataverse 29 . Cooking characteristics In the baseline dataset, a majority of time spent cooking (across the dataset) employed three stone fires, with the remaining less than 20% of the time spent using charcoal or firewood stoves. After introduction of the firewood cookstoves to all households, over 95% of the overall cooking time was spent using the new stoves, with consequent reductions in use of three stone fires and charcoal stoves, now together constituting less than 5% of total cooking time 29 (Figure 2). There were significant differences in fuel use in the before and after phases, with maize cobs widely used (in all but three households) post-intervention ( Figure 3). This was linked to the timing of the harvest: whilst pre-and post-intervention periods occurred at a similar time of year, the post-intervention phase coincided with the immediate post-harvest period such that maize cobs were freely available in the village and tended to be used as fuel in preference to other available fuel types such as wood and charcoal 29 . Qualitative observations revealed how this change in fuel use also explained the increase in 'combustion hours' in the post-intervention dataset, with the inefficient burning of maize cobs extending cooking time, compared with firewood use. PM 2.5 concentrations before and after cookstove introduction Median overall PM 2.5 concentrations pre-and post-intervention were not significantly different: pre-and post-intervention medians and interquartile ranges (IQR) 11.8 μg/m 3 (IQR: 3.8 -44.4) and 9.9 μg/m 3 (IQR: 2.2 -46.5) respectively (corrected data shown in Figure 4, with dotted line to denote the WHOrecommended 24-hour upper limit (PM 2.5 concentration 15μg/m 3 ) 37 . Comparison of pre-and post-intervention medians grouped by participant number confirmed no significant difference between these concentrations (Wilcoxon V=95; p=0.70). Matching activity data to traces, we found that median and interquartile range values during cooking activity before and after cookstove introduction were not significantly different (median and IQR for cooking-related concentrations pre-and post-intervention 79.4 μg/m 3 (IQR: 21.5 -397.0) and 80.6 μg/m 3 (IQR: 36.3 -307.4) respectively; V=86; p=1.00. Median and IQR concentrations were above WHO-recommended 24-hour upper limits throughout (corrected data shown in Figure 5a). During periods of no identified combustion activity ('background'), there was a statistically significant reduction in median PM 2.5 concentrations after the introduction of stoves, from 8.5 μg/m 3 (IQR: 3.0 -21.4) to 4.6 μg/m 3 (IQR: 1.0 -12.7); V=123; p=0.03. This reduction brought more of the values below the WHO limits (corrected data shown in Figure 5b). Qualitative findings Cookstove use: Observations throughout the village supported the finding from the monitoring sample of high cookstove uptake rates. On walking through the village, we frequently found people cooking on the cookstoves and there was good evidence of cookstove use at households we passed. Almost all the cookstoves were blackened with cooking smoke, and they were often covered in maize meal flour, suggesting habitual use. Notably, where previously three stones were to be seen in and around almost every household, and often smouldering in the background before or after formal cooking episodes, these were now much less frequently seen. In some cases, the stones or bricks were seen to be discarded outside the yard. This was confirmed when raised in discussion with household members who, when asked where their three stone fires were, responded, "palibe (there are none), we threw them away". This finding, while frequent, was not universal, however. In discussion, a few residents mentioned using fires concurrently with their stoves if cooking had to be done quickly. In two households, women reported children (who were unused to the new stoves) using fires for cooking, and some women said that the stoves could not be used for very large amounts of food (for example when making "thobwa", a fermented maize drink, and for cooking during special occasions such as weddings and funerals), although others' accounts asserted the opposite view, confirming their use of the new stoves for these purposes. One reason for not using the new stoves which was raised during several discussions was that firewood was sometimes in low supply. This related to the season, where there was little firewood to be found on the ground and this was sometimes damp or wet. In this situation, some residents bought small bags of charcoal, using this on charcoal stoves for the necessary household cooking. The purchase of firewood was uncommon as this was sold in large bundles which required a larger amount of money, as compared with small bags of charcoal. Perceived benefits of cookstoves: In response to questions around why participants liked and used the new cookstoves, there were a range of responses, of which the most common was that the stoves saved firewood. Participants used the same fuel as they would have used on their three stone fires -maize cobs (and at times maize stalks) as well as wood -and many claimed that their stoves "uses less maize cobs or firewood than three stone fire". The stoves were thus felt to be cost saving. A fieldnote made during a conversation with a resident, which -when raised -resonated with many others, read: "(Female participant explained that) it saves firewood, so saves money too. Sometimes she has to buy firewood, money goes further when using (a firewood cookstove)". Variations on this, which were also commonly stated, were that the fire in the stoves was shielded by the wind, and that the stove "keeps the heat", thus allowing for ongoing cooking or bathwater warming, without the continuing use of fuel. The second most commonly noted benefit of the stoves was faster cooking time ("imafulumira"), with some also noting the stove heating up more quickly than the time taken by a fire. "Our relish is now cooked in 10 minutes -previously, with a three stone fire, it would take until after 12" Fewer residents raised the issue of smoke in discussing benefits. When asked specifically about smoke levels, opinions were split, with some feeling that the stoves produced more smoke, but others feeling that fires were worse. When discussing smoke levels, many people talked about fuel: "with wood, the firewood stove is better, even if using maize cobs, although with these there's more smoke than with wood" "Wet wood is smoky at first, then it dries and is better -there's no difference between the stove and three stone fire. I would still use the firewood stove with wet wood" It was noted that the benefit of not having to tend to the fire in the stove as much as a three stone fire (as it was protected from the wind) and being able to move the stove inside or outside, allowed them a degree of control over control their smoke exposures while cooking. This was supported by a quantitative finding of more cooking taking place outdoors in the post-intervention phase than pre-intervention 29 . Perceived disadvantages of cookstoves: The main issue raised with the cookstoves was that of breakage. We observed a number of stoves which had cracks in the sides already, although in most cases these stoves continued to be used. The cracks rarely prohibited the use of stoves but did mean that these participants refrained from using very large pots on the stoves, out of caution, and from moving them to different places. We came across a few stoves in which, over time, cracks had progressed to significant breakage (and a piece of the stove was completely displaced). In one of these cases, the resident had bound wire around the cookstove rim to hold it together, allowing her to continue to use the stove. In other cases, the stoves could no longer be used and were discarded, with residents in these households having reverted to the use of three stone fires. When asked about replacing the broken stoves, residents were positive, with most stating that they would pay between 1000 and 2000MK (approximately 1.20 -2.50 USD): approximately the market price of the stoves. The extract from a conversation below illustrates many residents' thoughts on replacing the stoves: The main concern for most was that the stoves were not available for sale in the area, and that transport to the nearest market where they could be purchased would make their replacement unaffordable. Discussion Three to five months after the introduction of locally made clay stoves in the village, the new stoves were being used in most households, and for most of the cooking and bathwater warming activity. In the sample of participants involved in personal exposure monitoring, there was no change in PM 2.5 exposures with the introduction of the new stoves, although 'background' exposures -in the absence of specific combustion activity -were lower post-intervention. Qualitative data revealed a widespread approval of the stoves amongst residents, with the main reason stated being their more efficient use of fuel. Cracking of the stoves with use was a key issue raised, and is a relatively commonly reported issue with these basic stoves, often related to quality of clay or manufacturing processes 38,39 , although residents seemed keen to replace the stoves, should they be available for sale. The widespread use of the new stoves was apparent in both the time-activity data collected alongside air quality monitoring, and in participant observation data, with both sources clearly indicating a replacement of previous cooking methods with the new stoves. This is notable, given the prevalence of 'stacking' (combined use of multiple cooking modalities, old and new, rather than replacement) following the introduction of 'improved' cooking technologies [40][41][42][43] . This relates to the reasons for continued use of traditional stoves, which vary but include limitations of newly introduced technologies, need for concurrent cooking on multiple stoves, and fuel access and cost, as well as (less commonly) different context-specific cooking needs 40,41,43,44 . Participants in this study raised some of these issues, namely that of using multiple devices concurrently, although when asked they stated that they would use two stoves if they were available. Issues with fuel access were also sometimes raised, in keeping with previous findings around resource limitations in this setting 4 . In spite of the widespread cookstove use amongst the cohort, there was no difference in individuals' PM 2.5 exposures, either overall or during cooking periods, after introduction of the stoves. This is perhaps unsurprising given the lack of clear evidence of exposure reduction with these basic cookstove types, compared with traditional cooking fires 45 . Participants' observations of faster cooking time and less need to tend the fire when cooking on the new stoves signpost the potential for reductions in personal emissions on a larger scale -although this was not seen in our small sample of participants. Our finding of reductions in 'background' exposure (during non-cooking time) could reflect a previously reported greater reduction in smouldering emissions 46 This possibility is supported by the fact that frequent observations of household fires being left to smoulder in the pre-intervention period were greatly reduced in the post-intervention period when most of the fires were replaced by the more-efficiently burning cookstoves. Given the decrease further below WHO-recommended thresholds, this may be an encouraging direction of change from traditional stoves. These outcomes could be framed in terms of implementation science frameworks such as the RE-AIM framework 47,48 with statements relating to the high levels of 'adoption' and 'reach', poorer 'effectiveness' outcomes -judged in terms of researcher plans to reduce air pollution -and thoughts around ensuring 'maintenance' of the intervention in the longer term. This approach, with assessments made only in respect to researchers' predetermined aims and outcomes, was not the aim of the study however. Our ethnographic work allowed insights into participants' lived experiences, enriching the evaluation and helping us to understand it's value from a range of perspectives. In qualitative discussions, residents' main comments on the new stoves related not to 'smoke', but to perceived reductions in fuel use compared with three stone fires which they replaced, reflecting improvements in burning efficiency. This efficiency benefit is reported in the literature, although improvements with basic stoves tend to be modest compared with more advanced cookstoves 45,49,50 . The positive reception to the stoves seen in our study echo community responses to the introduction of the Jambar (another simple biomass stove with efficiency benefits) in rural Senegal 51,52 . Researchers Jeuland et al. note that "reducing firewood and charcoal consumption are important objectives in themselves -both from environmental and poverty alleviation perspectives" 22 . This is particularly relevant in a setting such as rural Malawi in which many residents' lives are shaped by severe economic scarcity, and where access to food, and fuel on which to cook daily meals, are prime concerns 4 . Researchers conducting the trial in Senegal and others have noted that participants' willingness to pay for new stoves was high despite their initial free provision, and that their widespread provision to all community members positively influenced their uptake 16,52 . Findings of the current study agree with this, in that positive reports of the stoves were far more forthcoming after their introduction across the village than before the intervention from the few households which owned the stoves 4 . This village-level approach is also important in view of the shared nature of air pollution, with widespread uptake of cleaner technologies required to accrue air quality benefits 53,54 . Strengths of our study lie in the combined use of qualitative observations and quantitative data collection to allow a realist evaluation of the intervention -delivered on a whole-village level -in its intended context, and activity matched exposure data. We acknowledge that our study had limitations, namely the small sample of participants involved in the quantitative 'air quality monitoring' component, and the slight difference in timing of pre-and post-intervention phases resulting in the widespread use of maize cobs as fuel in the post-intervention phase. Outcomes of air quality monitoring were broadly in keeping with expectations however, adding evidence around potential reductions in exposures during the 'smouldering' phase. These findings should be further explored with larger scale monitoring studies, using techniques such as those we have employed to decouple cooking-and non-cooking related exposures. In conclusion, whilst there were no cooking-associated reductions in PM 2.5 exposure after introduction of the cookstoves, the stoves were welcomed and widely used by residents across the village. Residents valued the efficiency and fast cooking of these stoves, as well as additional benefits such as a reduced need to tend the fire and the possibility of moving the site of cooking. Whilst significant improvements in air quality will require a more comprehensive approach 24,55,56 , accessible cooking solutions such as these stoves with the potential to meet communities' immediate needs represent a valued interim alternative to cooking on open fires. Scale up of production and distribution to allow more households to replace their stoves once broken, or even schemes to support local production, are required to allow more communities access to these simple technologies. Reflexivity statement The following reflexivity statement details key elements of the research partnership, conduct and reporting of the work presented above, in the hope that transparency with regard to transnational research practices will lay a foundation for more equitable ways of conducting collaborative research across the academic system. Study conceptualization 1. How does this study address local research and policy priorities? Air pollution is a global health priority. Malawi is a low-income country with high levels of air pollution and consequent morbidity. Cooking using solid fuels is thought to be a key contributor to airborne pollutant exposure in rural populations. Our interventional study -informed by an in-depth ethnographic account of air pollution (or 'smoke') in the setting -involved the introduction of a locally made cookstove in an effort to reduce individuals' exposures while also considering residents' other priorities relating to their health and wellbeing. How were local researchers involved in study design? The research assistant (HS) for this study is a local social scientist based in Malawi with previous experience doing research in this area. He was involved with study design and data collection and ensured that approaches and methods were context-appropriate throughout. The fieldworker (DM) is a resident in the village in which the study is based and contributed perspectives in study design and implementation as well as optimising linkages with the community throughout the wider study. Research management 3. How has funding been used to support the local research team(s)? Part of the research funding was used to provide salaries for local researchers -as above -and staff involved in the broader research grant including research governance and grants management. How are research staff who conducted data collection acknowledged? The research assistant and fieldworker worked with the main researcher on data collection, and the research assistant also supported data management activities. Both are authors of this paper with their specific contributions acknowledged appropriately. How have members of the research partnership been provided with access to study data? Study data is archived at Malawi Liverpool Wellcome Trust. Local researchers have direct access to the data. How were data used to develop analytical skills within the partnership? The PhD researcher (SS) supported the research assistant in quantitative data management as well as analysis of quantitative and qualitative data, helping to develop these skills further. Data interpretation 7. How have research partners collaborated in interpreting study data? Data interpretation involved discussions around analytical decisions and methods, which incorporated various members of the team (based in Malawi and the UK) How were research partners supported to develop writing skills? The lead author of this paper is a doctoral candidate. She led in writing the paper, with reflective input and advice from all partners. How will research products be shared to address local needs? Preliminary findings have been shared within the village at dissemination events. Earlier quantitative data have been presented at local research dissemination conferences and within the research institution (MLW), and these forms of sharing will continue with the present data. This manuscript will be made available to the wider global scientific community for discussion and development of the findings. How have early career researchers across the partnership been included within the authorship team? Please refer to question 8 above regarding leadership of the project. The study also incorporated a junior researcher in the LMIC setting as research assistant and a local fieldworker who had not previously had any research involvement, all included as authors. How has gender balance been addressed within the authorship? The research lead (whose doctoral work is represented here) is female, as are 3/7 of the authors, with representation from both local LMIC and HIC settings. Contributions to the study are acknowledged in the "Authors' contribution" section of the manuscript Training 13. How has the project contributed to training of LMIC researchers? The research assistant (HS) has been involved in the research process throughout, developing key skills, and has been supported in successfully applying for a Masters' scholarship in Global Health research. Involvement of the local fieldworker (DM) constituted her first experience of research participation. Both have significantly contributed to the project and are recognised accordingly in the authorship. These experiences will lay the foundation for further academic career development. How has the project contributed to improvements in local infrastructure? Whilst this is a small scale study, the project team have strived to support constructive engagement between the village community and the research institution throughout. Stoves were provided to all households as part of the study and links have been made with the local provider to enable residents to purchase replacement stoves in the future. Work is also underway to create a nursery/health centre in the village to express thanks to residents for their involvement and to provide continuity of employment for the local fieldworker. With reference to question 3 above, research governance, ethics and grant management systems of the local implementing partner (MLW) were supported through this grant. What safeguarding procedures were used to protect local study participants and researchers? The local ethics body and LSTM research ethics committee reviewed and approved the study protocol ensuring that both participants and researchers are protected throughout the study. Among other considerations, participants provided informed consent prior to their participation and, specifically, a named safeguarding lead (SS) was in place throughout, with various avenues of contact for participants to report any concerns, and structures for appropriate referral of any such reports. Tamire M, Addissie A, Skovbjerg S, et al.: Socio-Cultural Reasons and written and interesting paper that addresses just that issue. The Mbaula is indeed a cheap 'improved' cookstove in Malawi, one of the only improved stoves that communities can afford. Many communities have benefited from acquiring one over the three stone fire for a number of reasons, but for a number of reasons, the ability of the stove to reduce emissions had varied between studies. From our (soon to be published) scoping review it is also the only improved cookstove on the market which is cheap enough and accessible to the poorest populations in Malawi and therefore exploring community perceptions alongside HAP reduction is very useful. This particular study found no significant difference in PM2.5 levels between the three stone fire and the Mbaula, but it may be worth also examining the evidence regarding whether the Mbaula can reduce HAP in a laboratory setting through the Clean Cookstoves Catalogue and to discuss some of the differences in performance between laboratory and field setting use and how this might be addressed. http://catalog.cleancookstoves.org/stoves/385 Of particular interest is the discussion about the community views of the stoves and how this can potentially impact uptake and sustained use, as historically, there has been a greater focus on the technical aspects of the stove with less emphasis given to community preference. Might it be feasible to recommend the adoption of other behaviours that would reduce HAP alongside the Mbaula such as improved ventilation, facilities to dry wood, etc. as a means of reducing emissions further? If applicable, is the statistical analysis and its interpretation appropriate? Yes Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Yes Competing Interests: No competing interests were disclosed. I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard. Author Response 16 Mar 2022 Sepeedeh Saleh, Liverpool School of Tropical Medicine, Liverpool, UK Many thanks to Dr Stanistreet for this review and the interesting and very relevant comments. We eagerly anticipate reading the upcoming scoping review mentioned, and we agree with the suggestion of the need for further evidence around the stove's performance in different environments, in view of its wide accessibility in very low-income settings. We are grateful for the comments on the possibility of other 'behavioural' elements influencing overall exposures. A paper describing details of individual exposures in the village (pre-intervention), which provides evidence on the roles of fuel type and ventilation in shaping exposures, is currently under review, and we look forward to contributing further to these ongoing discussions around determinants of exposure in the rural African setting. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Marc Jeuland Sanford School of Public Policy, Duke Global Health Institute, Duke University, Durham, NC, USA Summary This is an interesting mixed methods study that explores adoption and use of simple biomass improved cookstoves in a rural village in Malawi. The authors conducted ethnographic research alongside measurement of exposures and time spent cooking to obtain richer insights on fuel and stove use practices and perceptions. The real strength of the study, in my opinion, is the ethnographic work, and I feel that the exposure results are somewhat overemphasized given the lack of a theoretical basis for expecting much exposure reduction. In trying to explain the exposure results (or lack thereof) the authors mostly speculate, without really nailing things down. More general and specific comments follow below. General comments I generally like this paper and the point it is making, but I think that the central role afforded to exposure actually detracts. It seems like the paper is almost apologetic about the lack of significant improvements in exposure. In so doing, it ironically elevates that issue and puts it on an equal footing with the more interesting ethnographic aspects and insights. I would urge the authors to re-equilibrate the discussion to be more consistent with their points about livelihoods. Introduction, the sentence: "Centring local perspectives in this way, as well as constituting arguably the 'right' approach to global health problems, can optimise the suitability and sustainability of any subsequent solutions". This phrase is needlessly ambiguous. First, I am not sure what the authors mean by the "right" approach. I also think "optimize the suitability and sustainability" is too strong a phrase. Just centering local perspectives does not guarantee optimizing these rather difficult objectives. Moreover, local perspectives can actually be in direct conflict with sustainability, if we are considering environmental sustainability, for example. ○ Study setting: I appreciate that the paper reference previous studies in this particular location, but a reader of this article would like to know a bit about the site without having to refer to those prior studies. What are the main economic livelihoods activities in this location? Are there particularities of cooking and fuel collection activities there that are worth highlighting? Etc. This is especially valuable for understanding the context and broader implications of this research. Study design and intervention: "Although they were known about by many in the village, few households already owned one of these firewood cookstoves." This is a very important sentence, that raises many questions. If known about, why didn't more households have these stoves? What had been the experiences with the stoves? Surely this would influence how people responded to the intervention. I noted that the authors discuss this point a bit on p.9 when emphasizing the importance of a "village-level" approach, but more background and discussion would be helpful about prior experiences and impressions. Data collection: It would be useful to understand seasonal aspects of data collection. I gather from the manuscript that the pre-and post-intervention observations were 3 months apart. Were these in significantly different seasons? Also, what was the season for the the ten week period of observations, and how does this relate to behaviors and insights obtained? (Note that this comes up on p.6 as an explanation for changed fuel use, but a reader wants to know such details much earlier). Results: Can the authors explain more why "combustion hours" increased post intervention? (p.5). Is this likely related to the intervention or due to some other time-varying phenomenon that was correlated with the intervention? ○ Figure 4 and the statement about medians pre and post intervention do not seem to match. Am I misreading the plot? ○ Did any households purchase firewood (not just charcoal)? My experience in Malawi tells me that this can be common in some areas, especially in the rainy season when firewood is wet. More details would be helpful here, including more explanation of the observation on p.7 that "This related to the season, where there was little firewood to be found on the ground ○ willingness to pay for new stoves was high despite their initial free provision, and that their widespread provision to all community members positively their uptake". There's a word missing in the latter part of the sentence. Is the work clearly and accurately presented and does it cite the current literature? Yes Is the study design appropriate and is the work technically sound? Yes If applicable, is the statistical analysis and its interpretation appropriate? Partly Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Partly sustainability" is too strong a phrase. Just centering local perspectives does not guarantee optimizing these rather difficult objectives. Moreover, local perspectives can actually be in direct conflict with sustainability, if we are considering environmental sustainability, for example. Thank you for this comment. The section in question has now been rewritten to improve the clarity and accuracy of the message. Thank you for picking up this valuable area of discussion. There was a lack of clarity in the paper as it was written, in relation to study timescales. This has now been revised (under the 'Study design and intervention' and 'Data collection' sections). The pre-intervention and post-intervention data collection periods actually took place during similar months over successive years. Most seasonal aspects should therefore be comparable between the two periods. A key difference here relates to the timing of the harvest, which fell during the post-intervention monitoring period (but not during the pre-intervention period of monitoring). This had repercussions on fuel use, as described in the paper. Results: Can the authors explain more why "combustion hours" increased post intervention? (p.5). Is this likely related to the intervention or due to some other time-varying phenomenon that was correlated with the intervention? ○ Thank you for raising this question. Our extended period of qualitative observation gave us insights into the likely reason for this, linked to the burning of maize cobs as fuel, which was inefficient and therefore extended cooking times, in comparison to firewood. This had been added to the text (under 'Cooking characteristics'). We are grateful for this astute observation. The data as presented in figure 4 reflected the original values with some zero values removed by the system, due to the log scale. This was the cause of the incoherent figure. All three box plots have now been replaced, with the use of corrected values to ensure the full data are presented. The text in the relevant sections has also been revised to reflect this. Did any households purchase firewood (not just charcoal)? My experience in Malawi tells me that this can be common in some areas, especially in the rainy season when firewood is wet. More details would be helpful here, including more explanation of the observation on p.7 that "This related to the season, where there was little firewood to be found on the ground and this was sometimes damp or wet." ○ Thank you for this observation. More information on fuel purchase has now been added, under 'Study setting and population', as well as the section described above. In relation to the specific question raised, we observed that village residents rarely bought firewood (this tended to be done only if households came across more money than they usually had access to) -mainly due to the costs of purchasing (large) bundles of firewood. Charcoalbeing more commonly sold in small bags -was the more usual purchased fuel of choice where firewood was not freely available. We are grateful, Prof Jeuland, for your raising of this question and for noting the confusion around the use of the word 'baseline'. The use of 'baseline' has been reviewed throughout the paper, and changes made, in line with reviewer suggestions, to avoid confusion. Regarding the question of the reduction in background exposures, the suggestion of smouldering is a proposed cause, but one supported by our extended observations in the field, both over a period of nine months before any intervention and following the introduction of the stoves. We regularly witnessed (and experienced) fires smouldering for periods of time following one cooking episode, and between episodes of cooking or bathwater warming, while household members served food, ate, and relaxed in the yard or veranda. During our time in and around households after the intervention, we noted that this was barely ever the case with cookstoves, and thus could be a likely mechanism for the reduced background exposures post-intervention. These details have been added to the manuscript in the relevant sections. As discussed above, there were few seasonal differences between the pre-and postintervention periods, as these periods covered similar months, both in the second half of the rainy season. Findings of significant breakage just a few months after intervention are concerning, but the authors don't really reflect much on this issue and how villagers were thinking about it. ○ Thank you for this point. Whilst the cracks in most cases did not constitute 'significant breakage' (most cracks were cosmetic, as described in the text, not interfering with cooking function), we agree that this had potential repercussions for intervention sustainability. This could be compounded by the issue of the stoves not being sold in or close to the village, and subsequent added costs of accessing now stoves. Unfortunately, these ongoing issues were too expansive to include in the current paper, whose scope was already felt to be quite broad. We agree that sustainability of this intervention over time will be an important subject to assess in future publications. Discussion: "Four to six months". I thought the observations began after 3 months. Or should it be 4 months? ○ Many thanks again for noting the confusion with regard to timescales in this paper. The 'Study design and intervention' and 'Data collection' sections have been revised to clarify these issues, as has this sentence. "Residents valued the efficiency and fast cooking of these stoves -responding to key local ○ priorities -and these factors as well as less need to tend the fire and the possibility of moving the site of cooking also hold the potential for small reductions in population-level exposure." The second part of this sentence goes too far, given the results of the study. And again, it almost devalues the first part of the sentence, apologizing for the lack of exposure improvements. We are grateful for this comment and agree with your point, Prof Jeuland. The sentence in question has been altered accordingly. Specific comments Introduction: "In Malawi, which is largely rural, air pollution is a persisting problem". Do you mean "persistent"? ○ Thank you for this suggestion. The word has been changed to 'persistent'. "Thus, while individual household interventions will not be sufficient to achieve clinically impactful reductions in PM2.5 there may be benefits to community-level adoption of locally relevant cleaner stove types in low-income settings such as Malawi." Is the "communitylevel" qualifier needed here? Wouldn't there even perhaps be benefits from individual adoption? (e.g., fuel and time savings) ○ Thanks again for this comment, with which we agree. We have removed the qualifier and slightly altered the sentence to reflect this point. Data Collection: "Discussions were often based around cooking activities, partly because this was the activity families were most often engaged in when spending time around the household." It would be nice to define "cooking activities". In particular, I find it hard to believe that cooking was the activity with the most time spent, unless cooking is defined rather broadly to include cleaning, firewood collection and preparation, etc. Some more precision here would be helpful. Cooking hours themselves are likely significant, but rarely the dominant form of time use in settings such as this one. ○ We are grateful for this interesting observation. Our meaning here was that, in the postintervention period, during which we mostly spent time and around households, the activities we observed (and at times took part in) around the household were largely related to cooking. This is in contrast to wider daily activities such as working on farms, bringing water from the well, and going to the shops, which took place away from the household. We also agree that the phrase 'cooking and related activities' better reflects our meaning here. We have revised the text to better explain this. Discussion: "Researchers conducting the trial in Senegal and others have noted that participants' willingness to pay for new stoves was high despite their initial free provision, and that their widespread provision to all community members positively their uptake". There's a word missing in the latter part of the sentence.	2022-03-19T14:28:14.371Z	2022-03-17T00:00:00.000	{ "year": 2022, "sha1": "34951270daf83f044b29cf4dfe9a2026578f2b4e", "oa_license": "CCBY", "oa_url": "https://wellcomeopenresearch.org/articles/7-52/v2/pdf", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "19bc8b892c990e350a920fbdd72eee9a1bffcfb2", "s2fieldsofstudy": [ "Environmental Science" ], "extfieldsofstudy": [ "Medicine" ] }
16097189	pes2o/s2orc	v3-fos-license	An Epidemic Model on Small-World Networks and Ring Vaccination A modified version of susceptible-infected-recovered-susceptible (SIRS) model for the outbreaks of foot-and-mouth disease (FMD) is introduced. The model is defined on small-world networks, and a ring vaccination programme is included. This model can be a theoretical explanation for the nonlocal interactions in epidemic spreading. Ring vaccination is capable of eradicating FMD provided that the probability of infection is high enough. Also an analytical approximation for this model is studied. I. INTRODUCTION Currently, there are some outbreaks of foot-and-mouth disease (FMD) in some developing and European countries. FMD is very infectious for cattle and sheep that are essential source for human food. The outbreaks of FMD cause a high loss in the national income, because export markets are closed for a country once a disease appears in it. Generally animals do not have natural immunity against FMD. Also, FMD can be transmitted through contacts with infected animals, tools, food and aerosol. So once an infected animal is discovered, one can consider the whole farm is infected. There are some strategies to control an outbreak. Mass vaccination means vaccinating the whole population with a certain rate. But vaccination is expensive and a virus usually has different configurations that require different types of vaccination. Also immunity due to vaccination often vanishes within some months. Therefore, one has to concentrate in vaccinating animals in contact to an infected case. This is called ring vaccination [1,2]. It follows the following procedures: (1) Once an infected case is discovered, the animals at the infected farm are slaughtered. (2) Animals are vaccinated within a ring with a certain radius around the infected farm. (3) A region around the infected farm is closed off. Ring vaccination always cost less than mass vaccination. Also it has been proven that ring vaccination is more suitable for FMD than mass vaccination [2]. Generally animals live in farms or herds forming clusters. Once a farm is infected, its nearest neighbours become at a high risk of infection. But this does not mean the interaction is just locally. In the 1987/1988 outbreak that occurred near Hannover, Germany [2], first two infected farms were discovered. A mass vaccination programme had been applied. After 3 months, another 2 infected farms were discovered at 18 Km away from the initial outbreak. Some days later, a third infected case was observed at a distance of 8 Km from the initial outbreak. This ensures the presence of nonlocal interaction in the spreading of an outbreak. Regular graphs can display the clustering property only. On the other hand random graphs [3] can show the nonlocal interactions without clustering. Small-world networks (SWN) [4,5] are shown to combine both local and nonlocal interactions. Also the concept of SWN was applied to many different systems. The results are often closer to the real systems than using regular lattices [6][7][8]. A coupled map lattice (CML) [9] is a dynamical system with discrete time and space steps and continuous states. It is used to model systems that consists of coupled elements. These systems often display spatiotemporal chaotic behaviour. The concept of CML is used to explain some aspects in many systems in different fields like biology, economics, etc. [9]. There are many mathematical models were suggested to study the epidemic spreading in a population [10][11][12][13]. According to the health of each individual, the population is classified to susceptible (S), infected (I) or recovered (R), In this work, we begin by studying a simple SI model on SWN. Then a model describes the epidemic spreading of FMD is suggested. It is an SIRS model [10] defined on SWN and a ring vaccination programme is considered. Also an analytical approximated model for the epidemic spreading on SWN is generalized to the inhomogeneous case using some concepts of CML. II. SMALL-WORLD NETWORKS The concept of SWN [4,5] is proposed to describe social networks. The social networks are characterized by two main properties: clustering and small-world effect. Small-world effect means the average distance between any two vertices is too short in compared with the size of the lattice. Only SWN is shown to satisfy the two properties together [5]. It is a connected 1-dimensional lattice of size L, with periodic boundary conditions. Some randomly chosen vertices are joined by some shortcuts to randomly selected other lattice sites. Here we consider shortcuts with length k = 1. Let φ be the average number of shortcuts per bond in the lattice. Hence for large L, the probability that two random vertices are connected by a shortcut is ψ ≃ 2kφ/L. Naturally the critical concentration of this graph p c is smaller than that for the ring p c = 1, and this was derived in [11]. Motivated by a simple susceptible-infected (SI) model, the spread of an epidemic in a population with random susceptibility is studied on SWN. This work generalizes that of Moore and Newman [11]. Consider the population occupy the vertices of a SWN of size L. Each individual is assumed to have a random susceptibility 1 − p(i), such that p(i) ∈ (0, 1). The propagation rule is: It is relevant to estimate the number of infected persons in this model as a function of time [14]. Setting q = 1 and p(i) = 0 ∀ i = 1, 2, . . . , L]. Hence the number of infected individuals will grow initially as a sphere with surface Γ d t d−1 , where Γ 1 = 2, Γ 2 = 2π, Γ 3 = 4π and so on. This is called the primary sphere. The probability of finding a shortcut is 2φΓ d t d−1 per unit time. Once a shortcut is reached a secondary sphere is formed and so on. Hence the total number of infected cases is given by (1) Its solution is In one dimension, one has For t < 1, the number of infected individuals grow as a power law t d /d!. While for t > 1, it grows exponentially. The transition occurs at t = 1 i.e. at t = 2φΓ d (d − 1)!. This result has an important effect in vaccination policies. It implies that vaccination should be administered as early as possible and with the highest possible ability to avoid reaching the exponential phase. Also immunizing individuals with shortcuts is more efficient than immunizing ordinary individuals. III. NUMERICAL SIMULATIONS The susceptible-infected-recovered-susceptible (SIRS) model [10] is more closer to FMD. The transition between the states S, I and R occur according to the following rules: (i) (Infection): An S-individual having at least one infected neighbour becomes infected in the next time step with probability q. (ii) (Recovery): I-individuals are recovered by a rate q 3 . (iii) (Losing immunity): The recovered individual change to the S-class by a rate q 2 . As a mean field approximation, the time rate of S, I and R are These equations have two steady states: To get a positive value for I 2 , N must be greater than S 2 . Then the disease can be eradicated if N q q3 < 1. This mean field approximation ignores the spatial structure of the lattice, and is valid for the case of global interacting model. But in reality, an outbreak spreads locally with some nonlocal interactions. To combine both local and nonlocal interactions, we define this model on SWN. A 1-dimensional lattice with periodic boundary conditions is assumed. A fraction of 2φ% of its individuals is assumed to interact with a third individual (randomly chosen through the lattice), in addition to the local interaction with the nearest neighbours. Also a ring vaccination programme is included to control the disease propagation. So the following procedure is added to that of the SIRS model: (Ring vaccination): With a probability σ, an infected case and its nearest neighbours is changed to the R-state. The shortcuting neighbours are not included in this step, because they are usually unknowns. The simulations were carried out on a SWN with L = 5000 and φ = 0.05. The shortcuting neighbours are fixed beforehand. The values of q 2 = 0.2 and q 3 = 0.05 are fixed through the simulations. The value of q is varied q ∈ (0, 1) and the corresponding smallest σ sufficient to eradicate the disease is calculated. The results are averaged over 10 independent runs and given in Fig. (1). An analytical approximation for the model is given in Appendix A. IV. CONCLUSIONS Using SWN is better than using regular or random lattices in modelling the outbreaks of FMD. SWN describe both local and nonlocal interactions. For a simple SI model defined on SWN: Initially the number of infected individuals grows as a power law, then after a critical value of time, it grows exponentially. So vaccination should be adminstered as early as possible to avoid reaching the exponential growth. Ring vaccination is capable of eradicating FMD even for high infection probability. APPENDIX A: ANALYTICAL CALCULATIONS Here a model is presented that approximates the epidemic spread on SWN. This model has been used before in different contexts [12,13]. Now it is generalized to the inhomogeneous case. Consider n patches each one contains a certain number of individuals (say animals). In general these patches are not identical. Infection spreads from infected animals within the patch and due to those diffusing from other patches. Then the time rate of the number of infected individuals in the ith patch is given by: The first term represents the infection within the patch with a rate λ i . The second represents the effect of other patches both nearby and far away at a rate µ i . The recovery rate is represented by γ i . Since the effect of other patches (second term in Eq. (A1)) is significantly smaller than the first one we expect that µ i ≪ λ i . Now the disease is eradicated if y i = 0 ∀i = 1, 2, ..., n. So the stability of this solution is studied. The system (A1) is a kind of CML, so we begin by presenting general stability results for CML. Then it is applied to Eq. (A1). Typically a 1-dimensional CML is given by where t = 1, 2, ... and j = 1, 2, ..., n. Consider the following inhomogeneous steady state of the system Linearizing around this solution, the system (A3) becomes A useful result on the eigenvalues of the system (A5) is Gerschgorin theorem [15]. It states that the eigenvalues λ of a square matrix [a ij ] satisfy \|λ − a ii \| ≤ j =i a ij , hence Hence the steady state solution (A4) is stable if the following conditions are satisfied For the homogeneous case: Then the stability conditions reduce to \|f ′ (α)\| < 1. Condition (A7) shows how diffusion may stabilize the system, since if f ′ (α j ) > 1 and f ′ (α j±1 ) < 1. Then Eq. (A7) may be still valid for enough large coupling. The stability conditions for the homogeneous solution (A8) for the system (A2) are Generalizing to the 2-species CML, we get where f t j = f (x t j , y t j ), g t j = g(x t j , y t j ). Consider the homogeneous steady state x t j = x 0 , y t j = y 0 . Linearizing around it, by taking x t j = x 0 + ε t j , y t j = y 0 + η t j . Also define the doublet: Then the linearized equations can be written in the following form where ∂f ∂x ∂f ∂y ∂g ∂x ∂g ∂y at (x 0 , y 0 ). (A13) This system corresponds to the direct product matrix Thus the steady state of the system (A10) is stable if all the eigenvalues λ k of the matrix A satisfy \|λ k \| < 1, where The equations of a CML with delay [16] are given by Repeating the above analysis, we obtain that the steady states x t j = y t j = x 0 are stable if all the eigenvalues of the following 2 × 2 matrix B = B j + B j+1 e −2πik/n + B j−1 e 2πik/n satisfy \|λ j,k \| < 1, where The stability conditions for the inhomogeneous solution (A4) of the 2-species system (A10) is: For the delayed system (A16), the stability conditions of (A4) is: where f ′ j ≡ df /dx j . From the above results, it is straightforward to prove the following proposition: Proposition 1: If local components are stable, then the homogeneous steady state solution of the corresponding CML is stable. Returning to Eq. (A1), the stability of the zero solution is determined by the eigenvalues of the following matrix For the homogeneous case where all the patches are identical, the parameters are independent of i. Hence the matrix A is circulant and the largest eigenvalue is λ − γ + µ. Therefore if λ − γ < 0 and λ − γ + µ > 0, then the disease would have been eradicated locally, but the diffusion term may cause it to persist. This shows the importance of the diffusion term (the long range edges in the case of SWN). Considering the more realistic inhomogeneous case where the patches are different. Also taking the natural assumption that µ i ≪ λ i , the matrix A is studied keeping only up to the second order terms in µ. Thus the Routh-Hurwitz stability conditions are that all the following determinants are positive: The effect of ring vaccination is expected to reduce λ i not µ i .	2001-08-09T09:28:54.000Z	2001-08-09T00:00:00.000	{ "year": 2002, "sha1": "176348106862d10c5a2331629369380e7ee6aa9c", "oa_license": null, "oa_url": "http://arxiv.org/pdf/nlin/0108013", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "176348106862d10c5a2331629369380e7ee6aa9c", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Geography", "Physics", "Biology" ] }
118535800	pes2o/s2orc	v3-fos-license	Renormalization Group Flow in Scalar-Tensor Theories. II We study the UV behaviour of actions including integer powers of scalar curvature and even powers of scalar fields with Functional Renormalization Group techniques. We find UV fixed points where the gravitational couplings have non-trivial values while the matter ones are Gaussian. We prove several properties of the linearized flow at such a fixed point in arbitrary dimensions in the one-loop approximation and find recursive relations among the critical exponents. We illustrate these results in explicit calculations in $d=4$ for actions including up to four powers of scalar curvature and two powers of the scalar field. In this setting we notice that the same recursive properties among the critical exponents, which were proven at one-loop order, still hold, in such a way that the UV critical surface is found to be five dimensional. We then search for the same type of fixed point in a scalar theory with minimal coupling to gravity in $d=4$ including up to eight powers of scalar curvature. Assuming that the recursive properties of the critical exponents still hold, one would conclude that the UV critical surface of these theories is five dimensional. I. INTRODUCTION In [1] scalar-tensor theories were studied where the purely gravitational part was given by the Einstein-Hilbert action. Here we generalize those results by including higher curvature terms. The main aim of our analysis is to understand if gravity remains asymptotically safe [2,3] under the inclusion of some matter component. Results about the renormalizability of gravity can depend crucially on the inclusion of matter. Already in the first one-loop calculations [4,5] it was shown that pure gravity is one-loop renormalizable but becomes one-loop nonrenormalizable in the presence of matter. In the context of the search for asymptotic safety, it was shown in [7] that the position and even the existence of a nontrivial gravitational fixed point in the Einstein-Hilbert truncation is affected by the presence of minimally coupled matter fields. In [6] we showed that this is also true when higher derivative gravitational terms are present. In [1,8,9] the effect of gravity on scalar interactions was studied, assuming the Einstein-Hilbert action for the gravitational field. In [1,8] it was shown that a nontrivial fixed point exists, where the purely gravitational couplings are finite while those involving the scalar field vanish. This is called a "Gaussian matter fixed point" (GMFP). In the present paper we extend these results by considering an interacting scalar field coupled to a class of higher derivative gravity theories which had been studied previously in [6,10]. We ask whether the scalar matter contribution is able to alter the results of the purely gravitational part considerably. In [6] the addition of minimally coupled matter components to R 2 -gravity (including all possible curvature invariants up to quadratic order) showed that the nontrivial fixed point structure is maintained in that case. We will see that this is largely the case here too, but since we now consider interacting scalars we will find that the dimension of the critical surface increases. There are clearly many possible applications in cosmology. Early work in this direction has been done in [11], using the beta functions of pure gravity. Taking scalar fields into account could have significant renormalization group running effects in inflation. Without the necessity of asymptotic safety, in effective field theory calculations the beta functions derived here could be useful e.g. for inflation [12], or in models where the Higgs field is used as the inflaton field [13]. Applications in the IR are possible, for example along the lines of [14] or the much discussed modified theories of gravity with some action based on different functional forms of the Ricci scalar (see e. g. [15]). We mention that the appearance of a scalar field in the low energy description of gravity has also been stressed in [16]. For a FRGE-based approach to that issue see also [17]. As in [1] the analysis is based on a type of Wilsonian action Γ k called the "effective average action" depending on an external energy scale k which can be formally defined by introducing an IR suppression in the functional integral for the modes with momenta lower than k. This amounts to modifying the propagator, leaving the interactions untouched. Then one can obtain a Functional Renormalization Group Equation (FRGE) [18] for the dependence of Γ k on k, where t = log(k/k 0 ). Φ are all the fields present in the theory. STr is a generalized functional trace including a minus sign for fermionic variables and a factor 2 for complex variables. R k is the regulator that suppresses the contribution to the trace of fluctuations with momenta below k. As the effective average action contains information about all the couplings in the theory, the FRGE contains all the beta functions of the theory. In certain approximations one can use this equation to reproduce the one-loop beta functions, but in principle the information one can extract from it is nonperturbative, in the sense that it does not depend on the couplings being small. A Quantum Field Theory (QFT) is asymptotically safe if there exists a finite dimensional space of action functionals (called the ultraviolet critical surface) which in the continuum limit are attracted towards a Fixed Point (FP) of the Renormalization Group (RG) flow. For example, a free theory has vanishing beta functions, so it has a FP called the Gaussian FP. Perturbation theory describes a neighbourhood of this point. In a perturbatively renormalizable and asymptotically free QFT such as QCD, the UV critical surface is parameterized by the couplings that have positive or zero mass dimension. Such couplings are called "renormalizable" or "relevant". Asymptotic safety is a generalization of this behaviour outside the perturbative domain. That means that the couplings could become strong. The FRGE allows us to carry out calculations also in that regime. Whether gravity is indeed asymptotically safe cannot yet be fully answered. However, since the formulation of the FRGE by Wetterich [18], many results support this possibility in various approximations [17,19,20,21,22,23,24], for reviews see [25]. Since the first application to gravity the necessary tools have been developed to make the approximation schemes more reliable including more couplings and studying their UV behaviour. In the approximations so far used, gravity has a nontrivial fixed point with a finite dimensional UV critical surface as is consistent with the requirements of asymptotic safety. The most common approximation method is to expand the average effective action in derivatives and to truncate the expansion at some order. In the case of scalar theory the lowest order of this expansion is the local potential approximation (LPA), where one retains a standard kinetic term plus a generic potential. In the case of pure gravity, the derivative expansion involves operators that are powers of curvatures and derivatives thereof. This has been studied systematically up to terms with four derivatives in [21,22,23,24] and for a limited class of operators (namely powers of the scalar curvature) up to sixteen derivatives of the metric [6,10]. In the case of scalar tensor theories of gravity, one will have to expand both in derivatives of the metric and of the scalar field. In this paper we will study the generalization of the action considered in [1,7,8] and [10] of the form where S GF is a gauge fixing term to be specified below and S gh is the corresponding ghost action. This action can be seen as a generalization of the LPA where also terms with two or more derivatives of the metric are included. This paper is organized in the following way. In section II we will give the inverse propagators resulting from the action (2) which have to be inserted into the FRGE to obtain the beta functions. In section III we describe the general properties of the GMFP. It is divided into two sub-sections. In section III A we show that minimal couplings are self-consistent in the sense that when matter couplings are switched off then also their beta functions vanish. In section III B, we analyze the linearized RG flow around the GMFP. We find that the stability matrix has block diagonal form which allows us to calculate its eigenvalues and eigenvectors in a recursive way. In section IV A we illustrate the existence of the GMFP and the properties of the RG flow near the GMFP in specific truncations where scalar matter fields are coupled nonminimally to gravity, including operators with up to four powers of scalar curvature and quadratic in the scalar matter field. In section IV B we consider minimally coupled scalar tensor theory including operators up to eight powers of scalar curvature and determine the dimensionality of the UV critical surface. We conclude in section V. A. Second variations Starting from the action given in eq. (2), we expand F (φ 2 , R) in polynomial form in φ 2 and R as In order to evaluate the r.h.s. of eq. (1) we calculate the second functional derivatives of the functional given in eq. (2). These can be obtained by expanding the action to second order in the quantum fields around classical backgrounds g µν =ḡ µν + h µν and φ =φ + δφ, whereφ is constant. The gauge fixing action quadratic in h µν is chosen to be where χ ν =∇ µ h µν − 1+ρ d∇ ν h µ µ , G µν =ḡ µν α + β¯ ; α, β, and ρ are the gauge parameters, we denote¯ =∇ µ∇ µ . The gauge fixing action eq. (4) gives rise to a ghost action consisting of two parts, S gh = S c + S b . The first part S c arises from the usual Fadeev-Popov prodedure leading to the complex ghost fields C µ andC µ . It is given by The second part S b arises for β = 0 and comes from the exponentiation of a nontrivial determinant which requires the introduction of real anti-commuting fields b µ which are usually referred to as the third ghost fields [26], These terms are already quadratic in the quantum fields. Then the second variation of eq. (2) is given by where = ∇ µ ∇ µ and h = h µ µ . Since we will never have to deal with the original metric g µν and scalar field φ, in order to simplify the notation, in the preceding formula and everywhere else from now on we will remove the bars from the backgrounds. As explained in [19], the functional that obeys the FRGE (1) has a separate dependence on the background fieldḡ µν and on a "classical field" (g cl ) µν =ḡ µν + (h cl ) µν , where (h cl ) µν is the Legendre conjugate of the sources coupling linearly to (h cl ) µν . The same applies to the scalar field. In this paper, like in most of the literature on the subject, we will restrict ourselves to the case when (g cl ) µν =ḡ µν and φ cl =φ. From now on the notation g µν and φ will be used to denote equivalently the "classical fields" or the background fields. B. Decomposition In order to simplify the terms and partially diagonalize the kinetic operator, we perform a decomposition of h µν in tensor, vector, and scalar parts as in [6,10], where h T µν is the (spin 2) transverse and traceless part, ξ µ is the (spin 1) transverse vector component, σ and h are (spin 0) scalars. This decomposition allows an exact inversion of the second variation under the restriction to a spherical background. With that in mind, we work on a d-dimensional sphere. For the spin-2 part, the inverse propagator is where δ µν,ρσ = 1 2 (g µρ g νσ + g µσ g νρ ). For the spin-1 part it is The two spin-0 components of the metric, σ and h, mix with δφ resulting in an inverse propagator given by a symmetric 3 × 3 matrix S with the entries As discussed in more detail in [6], to match the trace-spectra of the Laplace-operator acting on h µν with those obtained for the constrained fields after the decomposition, the first eigenmode of the operator trace over the vector contribution and the first two eigenmodes of the operator trace over the σ contribution have to be omitted. The trace over the h and δφ components should be taken over the whole operator spectrum instead. To handle the mixing of the scalar components in an easy way, we subtract first the two first eigenmodes from the complete scalar contribution from the matrix S and then add the first two trace modes which should have been retained for h and δφ. This requires to take into account a further scalar matrix B formed by the components of h, φ and their mixing term. It is given by whose trace contribution to the FRGE will be calculated on the first two eigenmodes of the spectrum of the Laplacian. Again, in order to diagonalize the kinetic operators occurring in the ghost actions eqs. (5) and (6), we perform a decomposition of the ghost fields C µ ,C µ and b µ into transverse and longitudinal parts, with ∇ µC µT = 0, ∇ µ C T µ = 0 and ∇ µ b T µ = 0. After this decomposition, the inverse propagators for the vector and scalar components of the ghost and third ghost fields are C. Contributions by Jacobians The decomposition of h µν ,C µ , C µ , and b µ gives rise to nontrivial Jacobians in the path integral, given by These Jacobians can be absorbed by field redefinitions which however introduce terms which involve noninteger powers of the Laplacian. To avoid technical difficulties, we therefore prefer to exponentiate these Jacobians by the introduction of auxiliary anticommuting and commuting fields according to the sign of the exponent of the determinant, see also [6,10]. One has to take their contribution into account while writing the FRGE. III. THE GAUSSIAN MATTER FIXED POINT The running of V a (φ 2 ) is calculated from the FRGE as where (∂ t Γ k )[φ 2 , R] is obtained for various fields in an analogous way as in [6,10]. Rescaling all fields with respect to the cutoff scale k, we obtain the dimensionless quantitiesφ = k . These dimensionless quantities we can use to analyze the RG flow and its FP structure. From the running of V a (φ 2 ) one can calculate the running ofṼ a (φ 2 ) using where the last term is calculated using eq. (19). A FP is a solution of the infinite set of functional equations ∂ tṼa = 0 for a = 0, . . . , ∞. This means that, at the FP, for each a the functionṼ a (φ 2 ) is k-independent, or equivalently that each coefficient of its Taylor expansion is k-independent. Since we assume that eachṼ a is analytic it can be Taylor expanded aroundφ 2 = 0, and therefore for i = 0, . . . , ∞, where the superscript i denotes the i-th derivative with respect toφ 2 . A. Minimal matter coupling of gravity at the GMFP The existence of a Gaussian Matter Fixed Point (GMFP), where all the matter couplings approach zero for k → ∞ and only the purely gravitational couplings have nontrivial values, was observed for finite polynomial truncations in [8]. In [1], its existence was proven for effective average actions of the form The existence of a GMFP can be shown to hold for the more general class of effective average actions considered in this paper. By definition, a GMFP is a point whereṼ a areφ 2 -independent, i.e. for i = 1, . . . , ∞. In this subsection we will prove that with the ansatz in eq. (23) all the equations in (21) with i = 1, . . . , ∞ are identically satisfied, thus leaving only the equations with i = 0 to be solved. We will give numerical solutions of these remaining equations for a = 0, 1 . . . , 8 in section IV. Now we explicitly analyze the structure of ∂ t F related to the second variation of the effective average action given in eq. (2) for the various field components. The second variation for h T µν and ξ µ has the form where we denote z := − . The functional form for Γ (2) k T,V is motivated by eqs. (9) and (10) from which we notice that it depends on V a at most linearly, with coefficients being functions of z and R, which are denoted here by f (z, R) and f a (z, R). For the scalar part, the second variation has the form where a prime denotes derivative with respect to φ 2 . Again the functional form for Γ (2) k s is motivated by eq. (11) which clearly tells that entries S σσ , S σh , and S hh depend at most linearly on V a , while the entries S φσ and S φh are linear combinations of φ V ′ a . The coefficients in these linear combinations are functions of z and R denoted here by l ij (z, R) and g i a (z, R). For the ghost part the second variation has the form This can be verified from eqs. (14, 15, 16 and 17). We first consider the contributions from h T µν and ξ µ . Since for them the second variation has the form given by eq. (24), the modified inverse propagator P k := Γ (2) k + R k and the cutoff R k will have the functional form where we have simply replaced z by P k (z) := z + R k (z) to obtain the modified inverse propagator. R k (z) is a profile function which tends to k 2 for z → 0 and approaches zero rapidly for z > k 2 . The RG-time derivative of the cutoff R k in eq. (27) is Using eq. (28) in the FRGE one finds that the contributions from h T µν and ξ µ have the form This can be justified by noticing that ∂ t R k given by eq. (28) depends at most linearly on ∂ t V b . On the r.h.s. of the FRGE, ∂ t R k occurs in the numerator, while the denominator contains the modified inverse propagator given in eq. (27) which depends at most linearly on V a . So we find that the r.h.s of the FRGE depends at most linearly on ∂ t V a . The coefficients in front of ∂ t V a are functionals of V a and are denoted by H a (V c ) and H ab (V c ). The contributions from the ghost parts will be simpler. Since they do not depend on the potentials, they will only give a constant contribution to H a . The contributions from the scalars are more involved due to the matrix structure. The modified inverse scalar propagator is obtained by replacing all z with P k in eq. (25). The cutoff is constructed in the usual way by subtracting the inverse propagator from the modified inverse propagator. This cutoff can be written as Then the t derivative of the cutoff given in eq. (30) is The modified propagator for scalars is the matrix inverse of eq. (25) with z replaced by P k . It is given by where Adj (P s k ) denotes the adjoint of the matrix (P s k ) (the matrix of cofactors). The determinant is a functional depending only on V a , φ 2 V ′ a V ′ b , and 2 V ′ a + 4φ 2 V ′′ a . This can be easily derived from the modified inverse propagator obtained from eq. (25). All entries of the adjoint of P s k consist of cofactors, thus it has the form where each entry depends additionally on P k and R. In order to calculate the RG trace, we multiply (P s k ) −1 with ∂ t R s k and then take the matrix trace. Doing this we note that φV ′ a is either multiplied with another φV ′ a or it is multiplied with φ∂ t V ′ a . So the scalar contribution to the FRGE has the form The contributions from the transverse traceless tensor and transverse vector can also be combined in the above expression to write the full FRGE contribution in the same way as above. Then ∂ t F = R a ∂ t V a . After having calculated the structural form for the running of V a (φ 2 ), we use it to calculate the dimensionless beta functional using eq. (20), which gives Inserting eq. (21) in eq. (35) we get the fixed point equation The above equation is identically satisfied when we take its Taylor expansion aroundφ 2 = 0 and use eq. (23). For example, taking one derivative with respect toφ 2 gives Settingφ 2 = 0 and using the GMFP conditions, we see that the right hand side will be zero. One can take successive derivatives to verify that this property indeed holds when higher derivatives are taken. The only equation which is not automatically solved in this way is the one where we evaluate eq. (36) atφ 2 = 0 and use eq. (23). This is just the FP equation for an f (R) theory with a single minimally coupled scalar. We will solve these equations in section IV. B. Linearized Flow around the GMFP The attractivity properties of a FP are determined by the signs of the critical exponents defined to be minus the eigenvalues of the linearized flow matrix, the so-called stability matrix, at the FP. The eigenvectors corresponding to negative eigenvalues (positive critical exponent) span the UV critical surface. At the Gaussian FP the critical exponents are equal to the mass dimension of each coupling, so the relevant couplings are the ones that are powercounting renormalizable (or marginally renormalizable). In a pertubatively renormalizable theory they are usually finite in number. At the GMFP, the situation is more complicated as the eigenvalues being negative or positive do not correspond to couplings being relevant or irrelevant. In principle, at the GMFP the eigenvectors corresponding to negative eigenvalues get contributions from all the couplings present in the truncation, thus making it more difficult to find the fixed point action. Thus understanding the properties of the stability matrix around the GMFP becomes crucial. Therefore we now discuss the structure of the linearized flow around the GMFP. It is convenient to Taylor expand the potentials V a (φ 2 ) as where λ (a) 2i are the corresponding couplings with mass dimension d − 2a − i(d − 2). We are assuming a finite truncation with up to p powers of R, i.e. a going from 0 to p, and q powers of φ 2 . In practice it has been possible to deal with p ≤ 8; as we shall see, it is possible to understand the structure of the theory for any polynomial in φ 2 , so one could also let q → ∞. Rescaling these couplings with respect to the RG scale defines dimensionless couplings λ 2i and the corresponding beta functions β The stability matrix is defined as Using the above definitions, numerical results tell that the stability matrix M has the form where each entry is a (p + 1) × (p + 1) matrix of the form while p is the highest power of scalar curvature included in the action. It turns out that, The various nonzero entries follow the same relations that were observed in [1]. In d dimensions they are where Using the same arguments as in [1], one can prove the above properties starting from eq. (34) neglecting ∂ t V a and ∂ t V ′ a on the right hand side (corresponding to a one-loop approximation). Solving eq. (34) beyond that level would require solving a functional differential equation and would be beyond the scope of this paper. However, the results presented in the next section suggest that these relations should hold exactly. They are relations independent of the gauge choice, however the entries of M 00 and M 01 are gauge dependent. The physical nature of the relations among the eigenvalues can be understood from the difference between the GMFP and the Gaussian fixed point where also the gravitational couplings would vanish. At a Gaussian fixed point, the critical exponents are determined by the mass dimension of the couplings, and therefore are all spaced by d − 2. At the GMFP, the gravitational couplings lead to some corrections to the critical exponents, but the correction to all exponents is the same, such that the spacing remains equal to d − 2. These relations have important consequences. Because the stability matrix at the GMFP has the block diagonal structure given by eq. (40), its eigenvalues are just the eigenvalues of the diagonal blocks. Since the diagonal blocks are related by eqs. (43) As M 00 depends only on the couplings λ (a) 0 , it is enough to include only these couplings into the action to find all the eigenvalues of the stability matrix. Therefore, the results for minimally coupled scalar-tensor theory determine the eigenvalues of the nonminimally coupled scalar-tensor theory. In particular, if one has calculated the dimension of the UV critical surface of the minimally coupled theory, one can also predict the dimension of the UV critical surface of the nonminimally coupled theory. To find all the eigenvectors of the stability matrix it is necessary to know also M 01 . One can write the eigenvectors as v = (v 0 , v 1 , . . . , v q ) T where each v i is itself a p + 1 dimensional vector. Then the vector V 0 = (v 0 , 0, 0, . . . , 0) T is an eigenvector if v 0 is an eigenvector of M 00 which can be seen immediately by multiplying it with M . The eigenvectors of M with the above form are eigenvectors for the eigenvalues of M 00 and can therefore be completely determined by just using M 00 . Thus we note at this point that these eigenvectors are mixtures of gravitational couplings only, they do not contain any contribution from matter couplings. Now consider a vector of the form V 1 = (v ′ 0 , v 1 , 0, 0, . . . , 0) T . Acting on it with M , and demanding V 1 to be an eigenvector of M corresponding to some eigenvalue ρ (a) 2 , we obtain two relations, The second equation in (47) tells that v 1 is an eigenvector of M 11 . Now due to equations given in (43) and (46), we note that v 1 = v 0 . Determining v 1 will then determine also v ′ 0 . In the same way one can go on to determine the next eigenvector. We then demand it to be a eigenvector of M . That means it should satisfy One notices immediately that v 2 is the eigenvector of M 22 , and using equations in (43) and (46) we conclude that v 2 = v 0 . Other equations would determine v ′′ 0 and v ′ 1 . This process can be continued to find all the eigenvectors. We will now illustrate the validity of these results in various truncations with scalar fields coupled minimally and nonminimally to gravity. A. Nonminimally coupled scalar field From here on we proceed as in [10]. We choose the gauge α = 0, β → ∞, and ρ = 0. This simplifies the calculation considerably because with that choice several arguments in the FRGE cancel with each other. The cutoff operators are chosen so that the modified inverse propagator is identical to the inverse propagator except for the replacement of z = −∇ 2 by P k (z) = z + R k (z); we use exclusively the optimized cutoff functions R k (z) = (k 2 − z)θ(k 2 − z) [27]. Then knowledge of the heat kernel coefficients which contain at most R 4 taken from [28] is sufficient to calculate all the beta functions. A further benefit of this choice of cutoff is that the trace arguments will be polynomial in z. This simplifies the integrations in the trace evaluation and is done in closed form. Inserting everything into the FRGE and comparing the terms with equal powers of R and φ 2 on each side of the equation will give a system of algebraic equations for the beta functions of the couplingsλ We carried out the calculation for effective average actions including up to R 4 and up to φ 2 in each potential V a . Such truncations include at most ten couplings. We find that a GMFP does indeed exist for all these truncations. The nonvanishing fixed point values for various truncations are given in table I, the corresponding critical exponents (the negative of the eigenvalues of the stability matrix) in table II. From the critical exponents one realizes at once several features. Though we carry out the full FRGE calculation we find that in general the real parts of the critical exponents ϑ 6.495 -21.579 2 5.224 -16.197 1.834 3 6.454 -20.756 1.071 -6.474 4 6.354 -21.342 0.792 -6.807 -3.865 TABLE I: Nonvanishing couplings at the GMFP. The index p is the highest power of R included in the truncation. All values are multiplied by a factor 1000. .670 -3.593 -5.182 1.261 2.772 -0.330 -5.593 -7.182 2i , like the couplings, but the corresponding eigenvectors involve strong mixing, as discussed in the text. For each i, the first two critical exponents form a complex conjugate pair given by ϑ ′ 0 ± ϑ ′′ 0 i and ϑ ′ 2 ± ϑ ′′ 2 i. relations among the eigenvalues will also hold at the exact level. The qualitative and quantitative properties turn out to be very similar to those of the purely gravitational theory. The inclusion of only four couplings with a = 0, 1 and i = 0, 1 leads to four attractive directions. The complex critical exponents ϑ ′ 0 ± ϑ ′′ 0 i are expected from the experience with the Einstein-Hilbert truncation. The existence of a second pair of complex critical exponents ϑ ′ 2 ± ϑ ′′ 2 i follows from the relation between the eigenvalues given in eq. (46). These complex conjugate pairs occur also when higher scalar curvature terms are included. When one includes also R 2 couplings, one encounters large positive critical exponents as known from the calculations in pure gravity [6,10,21,23]. Using eq. (46) one concludes that one has to go up to power φ 20 before encountering a negative critical exponent, so the critical surface would be twelve dimensional. But this is a fluke of the R 2 truncations due to the anomalously large positive critical exponent. The situation quickly normalizes when one adds further powers of R. Including R 3 couplings, classically one would expect only three positive critical exponents as the classical mass dimensions of λ 2 , and λ 1 , are 4, 2, 0, −2, 2, 0, −2, and −4 respectively. Apparently, the FRGE calculation, which includes quantum corrections with large mixing between the various couplings, produces instead six positive critical exponents in the R 3 truncation. The critical exponent ϑ (2) 2 is however very close to zero in consistency with the eigenvalue shift in eq. (46). This tells us that the truncation with p = 3 has a six-dimensional UV critical surface for any i ≥ 1. With the inclusion of the coupling for the R 4 operator whose classical mass dimension is −4, one notices that 0 < ϑ (2) 0 < 2. Thus one would expect that including the coupling for the operator φ 2 R 4 with classical mass dimension −6, in consistency with eq. (46), the critical exponent ϑ (2) 2 would be negative, and the critical surface would be five dimensional. Indeed, the inclusion of those couplings does make ϑ (2) 2 negative, leading to five negative and five positive critical exponents. One can then say, using eq. (46) in the truncation p = 4, that for any i ≥ 1, the critical surface would be five dimensional. To illustrate our results we display here the stability matrix for the R 4 truncation. The entries in the upper left 5 × 5 block and in the lower right 5 × 5 block are the same except the ones on the diagonals which differ by two. The upper right block is M 01 , the lower left one contains only zero entries: The eigenvectors corresponding to the five positive critical exponents in the R 4 truncation are given by The first complex conjugate pair of eigenvectors corresponds to the complex conjugate pair of critical exponents ϑ ′ 0 ± ϑ ′′ 0 with values 3.2608 ± 2.7722i, while the second pair of complex conjugate eigenvectors corresponds to the complex conjugate pair of critical exponents ϑ ′ 2 ± ϑ ′′ 2 with values 1.2608 ± 2.7722i. The last eigenvector corresponds to the critical exponent ϑ (2) 0 = 1.6698. We note that the eigenvectors corresponding to the eigenvalues of M 00 , namely the first complex conjugate pair of eigenvectors and the last one, have the same structure as was described in the previous section, i.e. (v 0 , 0, 0, . . . , 0) T , where v 0 is determined by just using M 00 . We note that these eigenvectors do not get mixing from the matter couplings, but only from the purely gravitational couplings. Further more, if we look at the eigenvectors corresponding to the eigenvalues of M 11 , namely the second complex conjugate pair of eigenvectors in eq. (50), which has the form (v ′ 0 , v 1 , 0, . . . , 0) T , we clearly notice that v 1 = v 0 , as described in the previous section. B. Minimally coupled scalar field Having verified that the properties of the stability matrix proved at one-loop level do also hold in the exact calculation, we now analyze higher order curvature terms retaining only the couplingsλ (a) 0 corresponding to a truncation with a minimally coupled scalar field. Then one obtains the non-Gaussian fixed points and critical exponents given in tables III and IV. We analyze these results and use them to make predictions for the nonminimal truncation. One observes that the addition of the scalar fields alters the results for pure gravity in [6,10] only by a small amount. Just as there, the UV critical surface becomes at most three-dimensional, and fixed point values for the cosmological and the Newton constant remain very stable. It has to be remarked that for those two couplings the oscillation in the fixed point value after the introduction of the R 2 -term is not as strong as in pure gravity. Also the critical exponent obtained after the introduction of the R 2 -coupling becomes large, but not as large as in pure gravity. So the addition of the scalar field seems to have already a little stabilizing effect on the R 2 -truncation. The introduction of the R 4 and R 5 -couplings leads to a second complex conjugate pair of critical exponents as soon as both couplings are included. Now it is easy to analyze how the dimension of the UV critical surface changes under the introduction of nonminimal matter couplings. In general, if a critical exponent ϑ (a) 0 is negative then ϑ (a) 2i will also be negative for all i > 0. From table IV we see that ϑ (a) 0 < 0 for all a ≥ 3, thus all ϑ (a) 2i < 0 for all a ≥ 3 and i > 0. However, since 4 > ϑ ′ 0 > 2, using eq. (46) we conclude that 2 > ϑ ′ 2 > 0. This means that there are two more attractive directions. From table IV one sees however that 0 < ϑ (2) 0 < 2 as soon as R 4 is included, thus we do not obtain any other attractive directions. IV: Critical exponents for increasing number p of couplings included. The first two critical exponents are a complex conjugate pair of the form ϑ ′ ± ϑ ′′ i. The same is the case for the fourth and fifth critical exponent ϑ (4) 0 ± ϑ (5) 0 i. So compared to [6,10] where a three-dimensional UV critical surface was obtained for pure gravity, interactions with scalar matter lead to a five-dimensional UV critical surface. V. CONCLUSION We have shown that a Gaussian matter fixed point does exist also under the inclusion of higher order curvature terms and their coupling to scalar fields. We verified that the properties of the stability matrix proven only at one-loop level hold also in the exact calculations. We exploited these properties to show the relations between minimally and nonminimally coupled scalar-tensor theory. In particular, we were able to calculate the critical exponents for the nonminimal scalar tensor theory from those of the minimal one. The introduction of minimally coupled scalar matter fields gives only slight quantitative corrections to the fixed point properties of the purely gravitational theory. The critical exponents again seem to converge with the inclusion of more curvature terms. The minimally coupled theory produces three positive critical exponents. We derived that the additional critical exponents in the nonminimally coupled theory will be the ones of the minimal theory shifted by constant values. This produces two more positive critical exponents. From that we can conclude that, in four dimensions, the scalar-tensor theory based on an action polynomial in scalar curvature and in even powers of scalar field gives rise to a five-dimensional UV critical surface.	2009-12-14T18:50:41.000Z	2009-11-02T00:00:00.000	{ "year": 2009, "sha1": "a3e7ad95f6122639de9304f4af9455d5a3bdea36", "oa_license": null, "oa_url": "http://arxiv.org/pdf/0911.0394", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "a3e7ad95f6122639de9304f4af9455d5a3bdea36", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Physics" ] }
229931366	pes2o/s2orc	v3-fos-license	Osteoprotective Effects of Loganic Acid on Osteoblastic and Osteoclastic Cells and Osteoporosis-Induced Mice Osteoporosis is a common disease caused by an imbalance of processes between bone resorption by osteoclasts and bone formation by osteoblasts in postmenopausal women. The roots of Gentiana lutea L. (GL) are reported to have beneficial effects on various human diseases related to liver functions and gastrointestinal motility, as well as on arthritis. Here, we fractionated and isolated bioactive constituent(s) responsible for anti-osteoporotic effects of GL root extract. A single phytochemical compound, loganic acid, was identified as a candidate osteoprotective agent. Its anti-osteoporotic effects were examined in vitro and in vivo. Treatment with loganic acid significantly increased osteoblastic differentiation in preosteoblast MC3T3-E1 cells by promoting alkaline phosphatase activity and increasing mRNA expression levels of bone metabolic markers such as Alpl, Bglap, and Sp7. However, loganic acid inhibited osteoclast differentiation of primary-cultured monocytes derived from mouse bone marrow. For in vivo experiments, the effect of loganic acid on ovariectomized (OVX) mice was examined for 12 weeks. Loganic acid prevented OVX-induced bone mineral density loss and improved bone structural properties in osteoporotic model mice. These results suggest that loganic acid may be a potential therapeutic candidate for treatment of osteoporosis. Introduction Osteoporosis is a progressive skeletal disorder characterized by dysregulation of bone remodeling, resulting in systemic reduction of bone mass and a high risk of bone fracture [1]. In elderly or postmenopausal women, osteoporosis occurs frequently because of estrogen deficiency, which plays a critical role in bone homeostasis [2,3]. Deficiency of female hormones trigger an abnormal bone remodeling process between bone formation (osteoblasts) and resorption (osteoclasts) [4]. Osteoblasts are derived from mesenchymal stem cells and play a central role in the regulation of bone mineralization and formation [5,6]. Osteoclasts are members of the monocyte/macrophage lineage and are found on old bone surfaces that break down old or damaged bone cells [7]. Impaired regulation between bone formation and resorption leads to inappropriate bone remodeling processes with serious bone loss, resulting in osteopenia and osteoporosis [8,9]. Currently, pharmacological treatment of osteoporosis, especially in women during menopausal period, is focused on selective estrogen receptor modulators, bone resorption inhibitors, or stimulators of bone formation [10]. Long-term treatment is required for osteoporosis; however, some pharmacological medications still have limitations such as side effects and unmet needs [11][12][13]. Natural herbal plants have been widely used as modern alternative medicines for the treatment of various diseases due to fewer side effects and are appropriate for long-term use and multifactorial effects [14][15][16]. Specific single physiological compound(s) isolated from natural plants have therapeutic effects on various diseases such as inflammation, type 2 diabetes, and cancer [17][18][19]. Gentiana is a genus of flowering plants used as a traditional medicine for the treatment of inflammation, skin diseases, and fever [20,21]. The Gentiana lutea L. (GL) root has been reported to improve gastrointestinal motility and liver function [22][23][24]. In recent years, bioactive phytochemicals of GL, including iridoids (loganic acid), secoiridoids (gentiopicroside, sweroside, swertiamarin, and amarogentin), and xanthones (isogentisin) have been identified [25]. One of the bioactive compounds, gentiopicroside, promoted protective effects against bone-related diseases such as arthritis and osteoporosis in ovariectomized (OVX) mice [26,27]. However, the effects of other bioactive compound(s) isolated from GL have not yet been studied. In this study, we aimed to identify bioactive constituent(s) responsible for the antiosteoporotic effect of GL extract. We purified a phytochemical single compound, loganic acid, from the root extract of GL. In addition, we evaluated the effects of loganic acid on osteoblastic and osteoclastic cell differentiation and in an animal model with OVX-induced bone loss. A Bioactive Component for Promoting Osteoblast Differentiation Isolated from GL Root Extract To examine osteoprotective effects of Gentiana lutea L. (GL) root extract, we investigated alkaline phosphatase (ALP) activity in osteoblast cells and tartrate-resistant acid phosphatase (TRAP) activity in osteoclast cells. The results indicated that a 30% ethanol extract of GL root promoted ALP activity in MC3T3-E1 preosteoblast cells, whereas it inhibited TRAP activity in primary-cultured osteoclast cells (Supplementary Figure S1). A previous study has demonstrated that the GL root extract contains numerous bioactive components [28]. We screened bioactive component(s) isolated from GL extract for antiosteoporotic agents. Briefly, an extract prepared from roots of GL was fractionated using chloroform, ethyl acetate, and butanol solvents ( Figure 1). . Ethanol extract of GL (30%) was fractionated using chloroform, ethyl acetate, and butanol. The butanol fraction was further fractionated into five sub-fractions by octadecylsilyl (ODS) gel column chromatography and eluted with acetonitrile (ACN) solution containing 0.05% trifluoroacetic acid (TFA) in a step-gradient manner (10 to 40%). Finally, a single bioactive compound was obtained. fr: fraction, HPLC: high-performance liquid chromatography. Next, each fraction was assessed for (ALP) activity, which is a characteristic marker of bone metabolism for enhancing bone formation [29]. As significantly increased ALP activity was observed in the butanol fraction, we further fractionated the butanol fraction into five sub-fractions (GL-B-01 to GL-B-05) by column chromatography. GL-B-03 showed the most significant effect on osteoblastic activity of MC3T3-E1 cells. The constituents of the final purified bioactive fraction (GL-B-03) were analyzed by proton nuclear magnetic resonance ( 1 H-NMR) and carbon-13 nuclear magnetic resonance ( 13 C-NMR) ( Figure 2). Loganic Acid Promoted Osteoblast Differentiation in Preosteoblastic MC3T3-E1 Cells To confirm the effects of the purified bioactive compound, identified as loganic acid, isolated from the GL extract, on bone formation in vitro, we evaluated osteoblast differentiation in MC3T3-E1 cells. Preosteoblast cells were differentiated by adding ascorbic acid and β-glycerophosphate, and co-treated with loganic acid (2, 10, and 50 µM). Bone formation is stimulated by either increasing proliferation of osteoblastic lineage or inducing osteoblast differentiation [30,31]. Our results showed that treatment with loganic acid did not affect the viability of MC3T3-E1 cells during the differentiation period ( Figure 4A). However, loganic acid significantly increased ALP activity in a dose-dependent manner in ALP-positive cells ( Figure 4B,C), indicating that loganic acid promoted osteoblast differentiation. Next, we investigated the promotive effects of loganic acid on bone formation by assessing mRNA expression of bone-enhancing markers such as Alpl (alkaline phosphatase, ALP), Bglap (bone gamma-carboxyglutamic acid-producing protein, osteocalcin), and Sp7 (osterix). In the present study, 50 µM loganic acid treatment increased mRNA expression levels of differentiation-associated markers of osteoblasts such as Alpl, Bglap, and Sp7 ( Figure 5). These results suggested that loganic acid induced osteoblast differentiation through upregulation of osteoblastogenesis-associated genes. After induction of osteoblast differentiation by co-treatment with 50 µM loganic acid, the mRNA expression levels of Alpl, Bglap, and Sp7 were evaluated by qRT-PCR using targeted gene-specific primers, and the expressions were normalized using mouse Gapdh mRNA level. LA: loganic acid, Induction: osteoblast induction media-treated group. * p < 0.05 vs. Induction (Student's t-test). Loganic Acid Inhibited Primary-Cultured Osteoclast Cell Differentiation To confirm the effects of loganic acid on bone resorption, we investigated the differentiation of monocytes (preosteoblastic cells) derived from bone marrow of a seven-week-old mouse. Monocytes were successfully isolated and confirmed using fluorescence-activated cell sorting analysis (Supplementary Figure S2). The isolated monocytes were differentiated into osteoclasts by adding the receptor activator of nuclear factor kappa-B ligand (RANKL) and monocyte-colony stimulating factor (M-CSF). After induction of osteoclast differentiation, cells were co-treated with loganic acid (2, 10, and 50 µM) and osteoclast differentiation was assessed by TRAP activity/staining. In the present study, loganic acid reduced TRAP activity of primary-cultured osteoclasts in a dose-dependent manner without affecting the cellular proliferation of monocytes ( Figure 6A,B). In addition, stained TRAP-positive cells were decreased by treatment with loganic acid, compared to the induction group ( Figure 6C). These results indicated that loganic acid inhibited osteoclast differentiation. Oral Administration of Loganic Acid Prevented OVX-Induced Osteoporosis In Vivo Based on the in vitro results, we further assessed the anti-osteoporotic effect of loganic acid in OVX-induced osteoporosis mice. Eight-week-old OVX mice were either orally administered with different concentrations of loganic acid (2, 10, and 50 mg/kg/day) or subcutaneously injected with 17β-estradiol (E2, 0.03 µg) (positive control group) for prevention of OVX-induced osteoporosis [32]. After 12 weeks of treatment, BMD of the right femoral bone was assessed using a PIXI-mus bone densitometer, and micro-CT bone structures were scanned for analysis of trabecular properties, such as bone volume (BV/TV), number (Tb.N), trabecular thickness (Tb.Th), and spacing (Tb.Sp). As expected, treatment of mice (positive control group) with E2 for 12 weeks inhibited OVX-induced BMD loss and rescued bone structural properties such as BV/TV, Tb.N, Tb.Th, and Tb.Sp when compared to the OVX group. Similarly, treatment with loganic acid prevented osteoporotic BMD loss in mouse femoral bone ( Figure 7A). In addition, the quantitative parameters demonstrated that loganic acid enhanced protective effects against OVX-induced osteoporosis ( Figure 7B). Micro-CT images of the right femoral bone revealed that loganic acid improved OVXinduced bone structural properties ( Figure 7C). Taken together, these results suggested that loganic acid inhibited OVX-induced osteoporosis in vivo and can be considered as a therapeutic agent for the prevention of bone loss. Discussion This study describes the antiosteoporotic effects of loganic acid derived from Gentiana lutea L. (GL) extract on osteoblastic and osteoclastic cells in vitro and in ovariectomized (OVX)-induced osteoporosis mice in vivo. Antiosteoprotective effects of GL extract were assessed through increased alkaline phosphatase (ALP) activity in osteoblast cells and reduced tartrate-resistant acid phosphatase (TRAP) activity in osteoclast cells. In the present study, we identified a bioactive compound, loganic acid, which was isolated from GL extract. Loganic acid enhanced the differentiation of osteoblast in preosteoblastic MC3T3-E1 cells but prevented the differentiation of primary-cultured osteoclast cells. Finally, loganic acid was found to exert protective effects against OVX-induced osteoporosis in mice. GL extract comprises various bioactive components [28], and in this study, we showed that GL extract promotes bone formation using MC3T3-E1 preosteoblast cells. Loganic acid is one of the main compounds present in GL extract [28]. Several therapeutic effects of loganic acid have been demonstrated, including inhibitory effects on inflammation [33], anti-obesity effects [34], and preventive effects on diet-induced hypertriglyceridemia and atherosclerosis [35]. However, the protective effects of loganic acid on bone metabolic disorders have not been elucidated yet. New bone formation is induced upon the differentiation of osteoblasts derived from mesenchymal stem cells, which is regulated via the transcription factors, such as ALP, osteocalcin, and osterix [36]. During osteoblast differentiation, an increase in the expression of ALP and osteocalcin regulates osteoblast function and bone mineralization [37,38]. Additionally, osterix is a zinc finger protein that is essential for osteoblast differentiation and bone formation [39]. Loganic acid increases ALP activity through increasing the expression of Alpl, Bglap, and Sp7, indicating that loganic acid stimulates osteoblast differentiation through activating osteoblast-inducing genes. Bone remodeling is regulated via maintaining an equilibrium between osteoblasts (bone formation) and osteoclasts (bone resorption) [40]. Osteoporotic postmenopausal women with aggressive osteoclastogenesis are at a high risk of fragility fractures [41]. Osteoclasts are differentiated via the fusion of monocyte/macrophage lineage cells originating from hematopoietic stem cells [42]. TRAP is considered as a reliable biomarker for osteoclast differentiation associated with bone resorption and remodeling [43]. In the present study, loganic acid was found to rescue TRAP-positive cells and activity, suggesting that loganic acid inhibits osteoclast differentiation. Ovariectomy involves the removal of ovaries and is a well-known postmenopausal model, resulting in estrogen deficiency, which plays an important role in bone homeostasis [44]. Dysregulation of bone homeostasis triggers the loss of bone mineral density (BMD), leading to osteoporosis in mice and humans [45]. OVX-induced mice exhibit significantly reduced BMD and abnormal trabecular bone structural properties. Loganic acid treatment inhibited osteoporotic BMD loss and improved quantitative bone structural properties, such as BV/TV, Tb.N, Tb.Th, and Tb.Sp. Moreover, micro-CT images revealed that loganic acid enhanced OVX-induced bone microarchitecture. These data suggest that loganic acid prevents osteoporotic trabecular bone loss in OVX mice. However, there were several limitations in this study, such as it is still uncertain if loganic acid is the only component present in GL extract, which is responsible for the anti-osteoporotic effects. In this study, we examined OVX female mice; however, it is unclear whether the osteoprotective effects of loganic acid are sex-specific. Finally, detailed analyses of the anti-osteoporotic effects of loganic acid are required to be performed further, including the assessment of physiological bone metabolic status of bone formation and resorption in the femoral bone. Fractionation, Isolation, and Structure Identification of Loganic Acid from Gentiana lutea L. Extract GL root was dissolved in 30% ethanol for 24 h at room temperature and then dried under vacuum. The remaining aqueous solution was filtered, and then sequentially fractioned using chloroform, ethyl acetate (EtOAc), and butanol (BuOH) to obtain chloroform (102.1 mg), EtOAc (81.1 mg), and BuOH (452.7 mg) extracts, respectively. The antiosteoporotic effects of each fraction were assessed by estimating ALP activity in MC3T3-E1 preosteoblast cells. The BuOH fraction was further fractionated by column chromatography using octadecylsilyl (ODS) gel column chromatography and then eluted using a 25% methanol solution into five subfractions (GL-B-01 to GL-B-05). Next, the GL-B-03 fraction was subjected to RP-18 column chromatography and eluted with an acetonitrile solution (containing 0.05% trifluoroacetic acid) in a step-gradient manner (10-40%). Evaluation of ALP/TRAP Activity and Staining For ALP activity, the differentiated osteoblasts were harvested and lysed with 1 mmol/L Tris-HCl pH 8. Water-Soluble Tetrazolium (WST) Assay Cells were incubated in a 96-well plate overnight and co-treated with different concentrations of loganic acid (2, 10, and 50 µM). Cell viability was evaluated using the D-Plus™ CCK cell viability assay kit (Donginls; Seoul, Republic of Korea). WST reagent was added to each well and the plate was incubated for 4 h. Next, absorbance at 450 nm was measured using a microplate reader (BIO-RAD, Hercules, CA, USA). Quantitative Reverse-Transcription Polymerase Chain Reaction (qRT-PCR) Total RNA was isolated from cultured cells using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. RNA was reversetranscribed into complementary DNA (cDNA) using the RevertAid™ H Minus First Strand cDNA Synthesis Kit (Fermentas, Hanover, NH, USA) with oligo (dT) 12-18 primers. PCR amplifications for osteoblastogenesis-associated genes were performed using the SYBR Green I qPCR Kit (TaKaRa, Shiga, Japan) and gene-specific primers. The specific primer sequences used in this study were as follows: forward 5 -CCA ACT CTT TTG TGC CAG AGA-3 and reverse 5 -TGA CAT TCT TGG CTA CAT TGG TG-3 for mouse Alpl; forward 5 -TAG TGA ACA GAC TCC GGC GCT A-3 and reverse 5 -TGT AGG CGG TCT TCA AGC CAT-3 for mouse Bglap; forward 5 -ATG GCG TCC TCT CTG CTT G-3 and reverse 5 -TGT AGG CGG TCT TCA AGC CAT-3 for mouse Sp7; and 5 -AGG TCG GTG TGA ACG GAT TTG-3 and 5 -TGT AGA CCA TGT AGT TGA GGT CA-3 for mouse Gapdh. By normalizing with mouse Gapdh, the relative gene expressions were determined as 2 −∆∆Ct , and fold-changes were determined by comparing with the untreated induction group. Examination of Bone Marrow Density (BMD) and Micro-CT Imaging in Mice Mice were anesthetized by intraperitoneal injection of tiletamine/zolazepam (Zoletil, Virbac Laboratories, Carros, France), and BMD of the right femur was measured using a PIXI-mus bone densitometer with on-board PIXI-mus software (GE Lunar, Madison, WI, USA). For micro-CT imaging, the right femoral bone was dissected and scanned by Inveon micro-CT (INVEON, SIEMENS, Munich, Germany) in Gyeonggido Business & Science Accelerator (GBSA, Suwon, Korea). Qualitative analysis of two-dimensional axial and 3D images was performed using CTvox software (BRUKER, Billerica, MA, USA). Statistical Analysis Data in the bar graphs are presented as mean ± standard error of the mean (SEM). All statistical analyses were performed using GraphPad Prism 6.0 software (GraphPad Software, San Diego, CA, USA). Statistical significance between two groups was calculated by Student's t-test. Comparison of multiple groups was performed by one-way analysis of variance (ANOVA) with Tukey's honest significant difference (HSD) post hoc test. A probability value (p) less than 0.05 was considered statistically significant. Conclusions In this study, we investigated osteoprotective effects of the phytochemical compound, loganic acid, isolated from GL extract. Loganic acid promoted osteoblast differentiation through upregulation of osteoblastogenesis-associated genes (Alpl, Bglap, and Sp7) and inhibited osteoclast differentiation. In vivo experiments performed in an osteoporotic mice model demonstrated that loganic acid treatment prevented OVX-induced bone loss and improved trabecular bone structure properties. Taken together, our results suggest that loganic acid may be a potential candidate to be used as an anti-osteoporotic agent.	2020-12-31T09:04:33.270Z	2020-12-28T00:00:00.000	{ "year": 2020, "sha1": "04fa51f0774644ca8b48fecc9fa19d4311b37f8d", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1422-0067/22/1/233/pdf", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "f1a06b98082fc1b5394178c6f6cc95e7a3125807", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
18477943	pes2o/s2orc	v3-fos-license	Breastfeeding as a Protective Effect Against Childhood Leukemia and Lymphoma Background Over the past several years, breastfeeding has been associated with many benefits as well as protective effects against many diseases. There is limited evidence for the relationship between breastfeeding and the incidence of leukemia. Objectives In this study, we evaluate the correlation of childhood leukemia and lymphoma with breastfeeding duration in children in southern Iran. Patients and methods Through this case control study, we compared 123 patients with leukemia and lymphoma to a control group of 137 healthy children. Statistical analysis was done using the Chi-square test and t-test as well as logistic regression methods. A P-value of less than 0.05 was considered significant. Results Our findings showed that breastfeeding duration had no significant difference between cases and controls. However, the rural living percentage in patients with leukemia and lymphoma was higher than in the control group (39.8% versus 14.6% [P < 0.001 and OR = 3.87]) and parents’ exposure to chemical materials during the war between Iran and Iraq was higher in sick patients (6.5% versus 0% [OR = 20.2%]). Conclusions The current study showed that breastfeeding duration has no protective effect against childhood leukemia and lymphoma. In addition, we suggest that some factors such as living in a rural area, smoking during pregnancy, parents’ exposure to chemical materials and low socioeconomic status can increase the incidence rate of childhood leukemia and lymphoma. Background Leukemia is the most prevalent childhood malignancy and represents 30% of all cancers in children (1,2). Acute lymphoblastic leukemia (ALL) is the most common type of childhood leukemia (3). Lymphoma is the third most common cancer in children living in America, and includes two important groups: Hodgkin lymphoma (HL) and non-Hodgkin lymphoma (NHL). The exact etiology of childhood leukemia and lymphoma has not been determined. However, several reports have shown that many risk factors including genetic abnormalities (such as Down's syndrome, Fanconi's anemia, ataxia telangiectasia and Bloom's syndrome), infections, radiation and exposure to some chemicals have been associated with these malignancies (4,5). In recent years, some studies have suggested that breastfeeding has a protective effect against leukemia and lymphoma in children as well as childhood infections. Indeed, human breast milk can boost the immune system as it contains many protective agents such as maternal antibodies, anti-inflammatory factors and immune cells that protect children from many infections and diseases (6)(7)(8). An infant's diet of breast milk can be effective against many diseases such as diabetes, cardio-vascular diseases, obesity, and even cancer (9). However, the relationship between breastfeeding and childhood leukemia and lymphoma is not clear. Objectives Although some studies suggest that breastfeeding has protective effects against Hodgkin lymphoma and other types of neoplasm such as ALL and AML, there are controversies among existing data about this relationship. In this study, we evaluate the correlation of childhood leukemia and lymphoma with breastfeeding duration in children in southern Iran. Patients and methods In this case control study we evaluated 134 patients with leukemia and lymphoma aged less than 18 years who were referred to an outpatient clinic or were hospitalized in the referral hospital in Shiraz, southern Iran from February 2011 to February 2012. The questionnaire included the following data: name, age, sex, residence, and leukocyte count at diagnosis, type of cancer, duration of breastfeeding or formula, tobacco consumption by parents during pregnancy, ABO and Rh blood groups, birth order, parents' exposure to chemical materials during the war between Iran and Iraq, and mother's education. All patients and parents signed the consent form and the project study was approved by the office of vice chancellor for research affairs of the university. Compiled data were gathered and analyzed by SPSS software (No.13) and statistical analysis was done by Chi-square and t-tests. In addition, a logistic regression model was made. A P-value of less than 0.05 was considered significant. Eleven patients were omitted from this study due to parents' inability to answer some questions. Results Two hundred and sixty children were studied. Of the children, 123 were patients with leukemia and lymphoma, and 137 were healthy children who served as the control group. The mean age of the experimental and control groups were 6.6 + 4 years and 6.8 + 4.2 years, respectively. For considering breastfeeding duration, we divided our patients into 4 study groups: 1) never breastfed, 2) breastfed for less than one month, 3) breastfed for one to six months, and 4) breastfed for six to twelve months. There was no significant difference between the patient and control groups (14.6% breastfed in patient group vs. 17.5% in control group in regards to six to twelve months of breastfeeding [P = 0.641]). The comparison of demographic parameters and other characteristics between the two groups is summarized in Table 1. Discussion Today, the exact etiology of childhood leukemia and lymphoma and the protective effect of breastfeeding against cancer are still unclear. Many epidemiological studies have suggested an association of breastfeeding with a reduced risk of lymphoblastic leukemia and other cancers (10,11). However, in a study done in 2011, Waly et al. (12) mentioned that breastfeeding duration did have not any protective effect against childhood ALL in Omani children. In contrast, findings from the large CCG epidemiologic studies of childhood AML and ALL showed a risk reduction among breastfed infants, particularly those breastfed for more than 6 months (13). A case study by Davis et al. (10) revealed that breastfeeding for more than 6 months had a protective effect for HL but not for AML and ALL. Conversely, a study by Martin et al. showed that breastfeeding slightly increased the incidence of ALL, HL and neuroblastoma. However, they mentioned that increasing breastfeeding duration can decrease the risk of ALL and lymphoma by about 5% (14). The results of the current study indicated that there was no significant relation between breastfeeding duration and childhood leukemia and lymphoma. Our findings show that a rural life increases the risk of childhood leukemia and lymphoma in comparison to an urban life; this factor has not been mentioned in previous studies, and it may be a reflection of lower socioeconomic levels in villages. The study conducted by Altinkaynak et al. (15). on leukemic children in Turkey showed similar results in that the mother being older (over 35 years old) and the mother's smoking habits were not risk factors of leukemia and lymphoma However, in several studies a mother's higher age (16) and smoking (17) have been shown as risk factors. As previously mentioned in this paper, low family income and low education levels for the mother increase the risk of leukemia in children. Also, children who are born later in the birth sequence (third, fourth or higher) are more susceptible to leukemia and lymphoma. Previous studies such as the one done by Perez-Saldivar et al. (18). On Mexican children also found the same results; it was indicated that the risk of leukemia increased in larger families (OR = 1.22). Altinkaynak et al. (15) conducted a study on Turkish children and concluded that the socioeconomic level in the leukemia patient group was lower than that of the control group (P < 0.001, which agrees with our results. However, there is controversy with this issue, and some researchers such as Bener et al. (19) have concluded that the mean family income is higher in the patient group. Through an ecological study in the USA in 2005, Buffler et al. (20) also concluded that a higher socioeconomic level is a risk factor for leukemia in childhood. Bener et al. (11) stated that a lower maternal age and low maternal education can increase the risk of ALL and HD. Notably, decreasing the duration of breastfeeding in educated mothers decreases the positive effect of this factor in their children. In the study of Saldivar et al. regarding the effect of parental smoking on Mexican children, it was reported that in children whose father had been in contact with carcinogenic ma-2 Iran Red Crescent Med J. 2016; 18(9):e29771. terials there was a higher risk of acute leukemia in childhood. For example, 55.4% of the sick children had fathers who smoked during their mother's pregnancy, but in the control group this percentage was 45.6% (OR = 1.5) (20). This result is similar to our findings. Regarding the relation between breastfeeding and childhood leukemia and lymphoma, the presence of confounding factors should be considered in future studies. In conclusion, the current study showed that breastfeeding duration has no protec-tive effect against childhood leukemia and lymphoma. In addition, we suggest that some factors such as rural living, smoking during pregnancy, parents' exposure to chemical materials during war and low socioeconomic status can increase the incidence rate of leukemia and lymphoma in children.	2018-04-03T03:51:19.721Z	2016-02-13T00:00:00.000	{ "year": 2016, "sha1": "5f98698833ec380139248409aa41a93c5d4ac4e1", "oa_license": "CCBYNC", "oa_url": "https://europepmc.org/articles/pmc5253205?pdf=render", "oa_status": "GREEN", "pdf_src": "PubMedCentral", "pdf_hash": "5f98698833ec380139248409aa41a93c5d4ac4e1", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
221221096	pes2o/s2orc	v3-fos-license	A simultaneous optical and electrical in-vitro neuronal recording system to evaluate microelectrode performance Objectives In this paper, we aim to detail the setup of a high spatio-temporal resolution, electrical recording system utilising planar microelectrode arrays with simultaneous optical imaging suitable for evaluating microelectrode performance with a proposed ′performance factor′ metric. Methods Techniques that would facilitate low noise electrical recordings were coupled with voltage sensitive dyes and neuronal activity was recorded both electrically via a customised amplification system and optically via a high speed CMOS camera. This technique was applied to characterise microelectrode recording performance of gold and poly(3,4-ethylenedioxythiophene)/polystyrene sulfonate (PEDOT/PSS) coated electrodes through traditional signal to noise (SNR) calculations as well as the proposed performance factor. Results Neuronal activity was simultaneously recorded using both electrical and optical techniques and this activity was confirmed via tetrodotoxin application to inhibit action potential firing. PEDOT/PSS outperformed gold using both measurements, however, the performance factor metric estimated a 3 fold improvement in signal transduction when compared to gold, whereas SNR estimated an 8 fold improvement when compared to gold. Conclusion The design and functionality of a system to record from neurons both electrically, through microelectrode arrays, and optically via voltage sensitive dyes was successfully achieved. Significance The high spatiotemporal resolution of both electrical and optical methods will allow for an array of applications such as improved detection of subthreshold synaptic events, validation of spike sorting algorithms and a provides a robust evaluation of extracellular microelectrode performance. Introduction The brain is one of the most intricate organs, functioning to control our physical senses, emotions and bodily processes. Specialised cells named neurons enable these complex functions through a network of connections and sophisticated electrochemical communication. The last decade has seen a surge of research which aims to interface these neural networks with electrode arrays in order to monitor and affect diseased pathways at an in vitro level through microelectrode arrays (MEAs) and an in vivo level through implantable electrode arrays [1]. The rationale behind this approach is that electrodes can correct or stimulate activity in certain neurons through a current pulse which causes depolarisation or hyperpolarisation of the cell [2]. Thus, achieving control over the diseased neural circuit. At the front-line of these devices are electrodes which interface with the neuron to record or stimulate activity. They are typically made of noble metal materials, such as platinum, but the demand for smaller electrodes to achieve high spatiotemporal resolution has strained the performance of these materials through a subsequent increase in impedance [3]. Therefore, this space has seen the rise of an exhaustive list of electrode materials which aim to increase the electrochemically active surface area whilst maintaining the same desired geometric area. Materials which have received the most attention due to superior electrochemical performance, stability and biocompatibility include poly(3,4-ethylenedioxythiophene) (PEDOT) [4][5][6][7][8], carbon nanotubes (CNTs) [9][10][11], glassy carbon [12], iridium oxide (IrOx) [13,14] and nanostructured platinum [13,15]. Characterisation methods of these electrode materials are well established and comprise morphological, biological, electrochemical and cellular activity recording investigations [8]. Morphological tests are generally carried out through scanning electron microscopy (SEM) to elicit the microstructure of the electrode material. Electrochemical tests comprise three individual parameters being (i) charge storage capacity (CSC c ) through cyclic voltammetry within the materials water window potentials, (ii) impedance through electrochemical impedance spectroscopy and (iii) charge injection limit through voltage transient measurements [16]. These electrochemical parameters predict favourable electrode properties for neuronal recording and stimulation. Biocompatibility assessment is carried out through the growth of cell cultures onto an MEA in vitro [8,17,18] and foreign body responses to implanted electrode arrays in vivo [10,13,19]. Recording and stimulation performance of the microelectrode are usually quantified through attribution of a signal to noise ratio (SNR) and the calculation of spikes arising from activated neurons in response to an injected current pulse. The SNR is an important metric and is generally calculated through (Eq 1) where S p−p is the amplitude of signals attributed to neuronal spiking and N p−p is the amplitude of electrical activity where no neuronal signals are recorded. The SNR is a generally accepted metric of electrode recording performance in vitro and in vivo. The impedance of a microelectrode is indirectly proportional to S p−p and directly proportional to N p−p , therefore if impedance is lowered a higher SNR can be achieved-justifying the SNR metrics place in characterisation of neuronal microelectrodes. However, a large determinant of S p−p is the distance of the signal source (neuron) from the microelectrode and this factor is not taken into account in the Eq 1. Therefore, if a cell is distant from the recording electrode the SNR may be reported falsely low for a specific electrode material. We hypothesise that a characterisation method which takes into account the distance of the signal source would more accurately represent the performance of an electrode material to transduce neuronal signals. The proposed metric will be named 'performance factor' and will require locating the firing neuron optically alongside simultaneous electrical recording of its activity. To do this, we aim to construct an electrophysiology system which can record electrical spiking activity from a population of primary neuronal cells, and couple this with optical imaging of neuronal action potentials through voltage sensitive dyes (VSDs). VSDs are optical indicators sensitive to membrane potential. They offer the possibility to visualise, in real time, the electrical activity of large neuronal populations with high spatial (up to 0.5 μm [20]) and temporal (μseconds) resolution [21,22], with the fastest VSDs based on electrochromic mechanisms (also known as the Stark effect) having less than 0.1 μs response times. Chien and Pine confirmed the ability of VSDs to detect hyperpolarisation events, action potentials and subthreshold synaptic potentials with simultaneous patch electrode recording [23,24]. Since then, improvements to VSD design have seen an increase in probe sensitivity from 1%/100 mV to 10%/100 mV fluorescence changes, leading to better detection of these electrophysiological events [20,25]. PLOS ONE In this paper, we present schematics for a low noise amplifier to acquire high quality electrical recordings from MEA devices as well as optimised VSD protocols and imaging techniques using a high speed CMOS camera. Primary hippocampal cells are cultured onto MEA devices which contain both bare gold and PEDOT coated microelectrodes to test the 0 performance factor 0 of different electrode materials. This metric is then compared to the conventional SNR calculation using the same materials. The constructed system is also validated for detection of neuronal signals from both visual and electrical sources through addition of tetrodotoxin (TTX), an irreversible sodium channel blocker. We believe that this methodology will more accurately characterise microelectrode performance, and provides a blueprint for further applications such as validation of spike sorting algorithms. Microelectrode array fabrication A custom 14 channel microelectrode array to allow for extracellular neuronal recordings was constructed. Gold slides with a titanium adhesion layer (Au/Ti) were purchased from Deposition Research Laboratories Incorporated (DRLI) and patterned using conventional photolithographic protocols. nLOF 2070 was used as a mask to allow for selective etching of Au and Ti layers, SU-8 2005 was then used as an overlying insulation layer-exposing 20 μm diameter areas of gold on each of the electrodes. Each microelectrode could be individually addressed through header pins which were soldered onto the MEA at a distant site. Microelectrode modification with conducting polymer. Conducting polymer (CP) modification was undertaken to increase electrochemical surface area, decrease impedance and theoretically improve microelectrode recording performance. Poly(3,4-ethylenedioxythiophene) doped with polystyrene sulfonate (PEDOT/PSS) was electrochemically polymerised onto gold microelectrodes from a solution of 0.01 M 3,4-ethylenedioxythiophene (Sigma, 483028) and 0.1 M poly(sodium 4-styrene sulfonate) (Sigma, 243051). Polymerisation was carried out in a three electrode cell consisting of a working electrode (gold microelectrode), silver/silver chloride (Ag/AgCl) reference electrode and gold counter electrode. Galvanostatic polymerisation (constant current) was employed with a current density of 2 mA cm −2 until 1000 nC of charge was passed (318 mC cm −2 ). This was carried out using the Biologic VSP-300 electrochemical workstation. Impedance characterisation. Microelectrode viability and electrochemical properties were assessed through electrochemical impedance spectroscopy (EIS). A three electrode set up was employed, consisting of an Ag/AgCl reference electrode, gold counter electrode and the microelectrode to be tested as the working electrode. A 10 mV sinusoidal wave was applied (Biologic VSP-300 electrochemical workstation) from 1 Hz to 10 kHz at open circuit potential in artificial cerebrospinal fluid (ACSF) to mimic impedance properties during recording. MEA acquisition system MEA to amplifier interface. Microelectrodes on the MEA were connected to the amplifier input using an interface board. The board plugged into the MEA header pins and sat over the MEA slide with a window allowing the microscope objective lens to access the solution. Outputs from the board consisted of 14x 2-pin headers, with each pair consisting of a pin tied to a microelectrode on the MEA (connected to non-inverting input at amplifier) and the other to a common Ag/AgCl reference immersed in solution (connected to inverting input). The board also had a ground plane which was connected to the amplifier ground to reduce noise in the measurements. Amplifier. The amplifier circuit (S1 File) was designed to be stable, low noise and low cost (US$21 for the first channel + US$11 for each additional channel). Full characterisation of this system can be found at [26]. Power circuitry consisted of a 9 V battery source (negative terminal connected to ground) with an ultra-low noise, linear regulator (LT3042, Linear Technology) used to maintain supply voltage to the circuit at +5 V. A reference voltage (REF) was set at 2.5 V by REF5025 (Texas Instruments), which provided a low noise, low drift reference potential. Two stages of amplification were used, the first stage being an instrumentation amplifier (LT1167, Linear Technology), and the second, an operational amplifier (LT1678, Linear Technology). Inverting and non-inverting input signals were AC coupled using with a high pass cut off frequency set at 0.7 Hz. To prevent voltage bias drifts at the input through capacitive charging, discharge paths were utilised using resistors. A LT1167 instrumentation amplifier was chosen at the first stage of amplification due to its low noise operation, high common mode rejection ratio (140 dB at a gain of 1000), low input bias current and high input impedance (200 GO) which allows the use of high impedance sources without additional offset voltage errors [27]. The high common mode rejection ratio (CMRR) ensures that the desired differential signal is amplified and unwanted common mode signals are attenuated. A single resistor sets the gain for the instrumentation amplifier at 1000. The common mode voltage is removed from the original signal by the instrumentation amplifier and results in a single-ended output voltage referenced to the voltage on the REF pin (2.5 V). The output voltage from LT1167 is high pass filtered with a cut off frequency set at 15.9 Hz and connected to the non-inverting input of LT1678 with reference to 2.5 V. Gain at LT1167 was set to 2 with a non-inverting feedback loop. Analog to digital converter. Amplified signals were digitised using a data acquisition (DAQ) device from National Instruments (USB-6356). Samples were acquired simultaneously at 20 kHz per channel with 16 bit resolution. A custom built LabVIEW programme (S2 File) was used to interface with the DAQ and acquire/log data which was saved in .tdms format for later processing. Cell culture Primary hippocampal neurons were obtained from P0 Wister rat pups under the University of Auckland' Animal Ethics Committee approval (AEC numbers 1504 & 2051). The devices were immersed in deionised (MilliQ, 18.2 MO.cm) water for a minimum of 24 hours prior to culture to promote neuronal growth in order to remove aqueous contaminants left on the surface of the MEA devices during microfabrication and polymerisation processes. Primary hippocampal neurons from Wistar rats were cultured at postnatal day zero using established techniques [28,29] onto the MEA devices. The MEA devices were first sterilised with 100% ethanol and exposure to UV light then coated with 10 μg mL The resulting suspension was plated onto the MEA device at a density of one hippocampi per MEA setup. The MEA devices were then placed in a 5% CO2 incubator at 37˚C for 28 days. Half the NBM was replaced at days in vitro 1 (DIV 1) and a quarter replaced at DIV 7, 14 and 21. Electrophysiological recording of neurons through the MEA was carried out in ACSF following VSD staining (detailed below). Image acquisition. A high speed CMOS camera (MotionPro X3, IDT) was fitted onto a fluorescent microscope (Leica DM RXA2) consisting of a 50 W mercury light source (Leica), an I3 filter cube and a x10 water immersion fluorescent lens (Leica 10X/0.3 HCX APO). Excitation/emission maxima for Di-4-ANEPPS were approximately 465/635 nm (as measured in model membranes), respectively-I3 filter cube was characterised by an excitation filter at 450-490 nm and a long pass emission filter for wavelengths over 515 nm, making it suitable for use with Di-4-ANEPPS. Note that results could be further optimised by using filters provided by the manufacturer of the VSD as spectra may be shifted in live cell experiments by up to 100 nm. The high speed camera was set at an acquisition rate of 1000 Hz with an exposure time of 958 μs (single exposure). Image acquisition and shutter position (open or closed) on the microscope were triggered by a digital signal from the DAQ to synchronise visual and electrical acquisition start times. Fluorescence intensity within the optical images was quantified using the Time Series Analyzer (V3) plug-in within ImageJ. A region of interest (ROI) around the visibly firing neuron was first selected, then the average fluorescence intensity was calculated within this ROI across the recorded frames. The ROI was saved and applied to subsequent data sets which contained the same neuron to ensure consistency within the analysis. MEA fabrication MEAs were successfully fabricated and uncoated gold microelectrodes displayed typical impedance spectra with a constant slope of -0.8 from the \|Z\| vs frequency bode plot (Fig 1), indicating capacitive impedance. This is consistent with previous reports of uncoated gold microelectrodes due to their capacitive mechanism of charge transfer at the electrode/electrolyte interface [31]. Electro-polymerisation of PEDOT/PSS onto the gold surface was accompanied by a drastic reduction of \|Z\| at all frequencies in the impedance spectra. This is commonly observed and can be attributed to the large electrochemical surface area offered by PEDOT/ PSS [4,32]. Electrodes intended for extracellular neuronal recordings are often characterised by their impedance magnitude at 1 kHz-a value close to the frequency of neuronal signals. In this case, PEDOT/PSS is dominated by its access resistance at 1 kHz therefore a comparison at a frequency of 100 Hz is more descriptive of the difference between the two electrodes. PEDOT/PSS produced a substantial drop in \|Z\| 100Hz from 4305 ± 342 kO to 67 ± 2 kO. Lower impedance magnitude is favourable for recording due a smaller noise floor and improved charge transfer properties [16,33]. MEA acquisition system The amplifier boards were assembled into an aluminum case to shield the inputs from electromagnetic noise, preventing saturation between the amplifier supply rails. Three-core wires, consisting of two insulated wires and a shield, were used between the MEA interface board and amplifier input. Although these wires successfully reduced noise at the inputs, it was PLOS ONE found that further reduction could be achieved by tying the cable shielding to the 2.5 V reference. The final intrinsic output noise of the amplifier was calculated to be 6.92 μV peak−peak (inverting and non-inverting input shorted). Peak to peak noise values rose to 38.2 ± 4.4μV or 19 ± 2.2μV when amplifier inputs were connected to gold or PEDOT/PSS microelectrodes, respectively. (average ± SD (n = 7)). This can be explained by high electrode impedance values, with gold having significantly higher values than PEDOT/PSS [16,33]. Insertion of the objective lens into the media bathing the electrodes caused the amplifier to saturate with noise. This was likely due to noise being introduced into the system by the microscope and was resolved by connecting a metal portion of the microscope chassis to amplifier ground (Fig 2). Cell culture Primary hippocampal neurons were cultured onto the MEA devices and displayed growth comparable to control cover slips routinely used in cell culture protocols. A time lapse of neuronal growth was achieved by taking photos at DIV 7, 14 and 21 to assess health and density (Fig 3). It can be seen that growth and connection density between neurons are consistent between the two samples. Fig 4A show the progression of an action potential as visualised by the VSD. Initially neurons were incubated with 1 μM Di-4-ANEPPS, however signal strength was poor and individual cells were difficult to visualise on-screen. Increasing the incubation concentration to 2 μM, resulted in clearer images with visible neuronal processes extending from the cell body. This was still within the manufacturers recommended loading concentration of 0.2-2 μM. Increases in the excitation intensity or exposure time to improve signal quality were avoided to reduce the occurrence of phototoxicity and maintain temporal resolution, respectively. It should also be noted that while this study employed Di-4-ANEPPS, newer PLOS ONE generation VSDs, such as di-2-ANEPEQ, di-3-ANEPPDHQ and di-4-AN(F)EPPTEA may have resulted in improved imaging [34]. Camera acquisition settings were adjusted to maximise the quality of digitised fluorescent signals. To visualise electrical activity high temporal resolution (<2 ms) and spatial resolution (<5 μm) was required. The final settings utilised an acquisition rate of 1000 Hz with an exposure time of 958 μs meaning a photo was taken every millisecond. The camera sensor gain was increased (x2) to accommodate for the decrease in signal strength due to low exposure times and 2x2 image binning was applied to further improve signal quality. Further binning at 3x3 and 4x4 did not provide any additional benefit and resulted in unnecessary loss of spatial resolution. The representative electrical recording trace shown is derived from one PEDOT/PSS microelectrode which is indicated on the images in Fig 4 as a red encircled dot. The recording was band-pass filtered between 200 Hz and 2000 Hz using a 4 th order butterworth IIR filter to isolate neuronal spike data. The spikes were easily distinguishable from the noise floor making action potential identification simple. A digital cue to initiate acquisition resulted in synchronised activity for both electrical and optical acquisition methods-with overlapping activity in both methods seen clearly from Fig 4A, 4B & 4C. Spikes marked with a star (?) were identified electrically through the microelectrode but not optically. This could be due to a different neuron firing at a distant location, not within the cameras field of view. This raises an issue as a large field of view is needed to make full use of VSD within this method. A way to remedy this problem is to decrease the objective lens magnification or adapt the set-up so that it can be used with a fluorescent macroscope capable of viewing the entire slide [35]. It should be noted that microelectrodes on MEAs are typically separated by a distance of around 25 to 50 μm, unlike the MEA presented here, therefore more than one electrode would typically be visible within the cameras field of view. The recorded electrical signals were confirmed to originate from neurons through use of TTX, an inhibitor of neuronal action potentials through binding of voltage-gated sodium channels (Fig 5A). TTX (1 μM) was added to the bath and recordings were taken at 1 minute, 5 minutes and 20 minutes to assess the effects of the toxin. The optical measurements displayed in Fig 5B indicate representative changes in fluorescence intensity waveforms of one The images were taken using differential interference contrast microscopy through a light microscope equipped with a 20x water immersion objective lens. Scale bar represents 100 μm. https://doi.org/10.1371/journal.pone.0237709.g003 PLOS ONE action potential before and after TTX addition. After 1 minute a significant reduction in electrical activity was noticed, optical imaging appeared to suffer from a slight loss in fluorescence near the electrode. After 5 minutes, almost no action potentials could be sensed through the microelectrodes, and optical signals have started to show clear signs of diminishing activity. 20 minutes post TTX addition revealed diminished optical signals where only activity within the cell body could be sensed which is most likely due to subthreshold activity [36]. The antagonism of neuronal activity in both optical and electrical recordings confirms neuronal signals in both these methods. These data also demonstrate the higher sensitivity of optical recording methods for subthreshold neuronal activity, where optical signals within the cell body are still prominent following TTX addition, whereas electrical recordings diminish significantly making identification of subthreshold events difficult. PLOS ONE Following the successful validation of this combined optical/electrical system, it was used to characterise the recording performance of gold and PEDOT/PSS microelectrodes. More specifically, a traditional SNR metric was attributed to each electrode material, followed by its assessment through our own performance factor measurement which takes into account the distance of the signal source. Fig 6 shows two images with neuronal activity next to a gold or PLOS ONE PEDOT/PSS microelectrode. The distance between neuron and electrode is clear and more robust calculations regarding electrode recording performance can be made through application of Eq 2 [37][38][39]. Eq 2 can be used to estimate the voltage at the microelectrode tip (V) at position x, y, z following an action potential, modeled as a transmembrane current source (I), at position x', y', z' (assuming an infinite volume conductor with homogenous extracellular electrical conductivity (σ)). Vðx; y; zÞ ¼ I 4ps 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 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 ffi ffi ffi ffi The expected potential at the electrode site can then be divided by the actual potential recorded and a 'performance factor' (i.e. recorded potential/expected potential) can be associated with the microelectrode used. The performance factors of gold and PEDOT/PSS electrodes were calculated to be 0.02 and 0.06, respectively (Table 1). This calculation was performed using only the x and y plane, where x,y were microelectrode coordinates and x',y' were coordinates of the neural process closest to the microelectrode. Solution conductivity was assumed to be 1.45 S m −1 [40], and a transmembrane current of 10 nA was used to model a neuronal action potential based on data from intracellular patch recordings-this value may differ and a more accurate calculation can be made by obtaining the real value for I. These results show that PEDOT/PSS microelectrodes are three-fold more effective in transducing biological currents, as predicted through impedance measurements. When comparing this method to SNR values as calculated through Eq 1 a large discrepancy between the results is noticed. SNR values were calculated to be 2.5 and 20 for gold and PEDOT/PSS, respectively. Traditionally, this would be reported as PEDOT/PSS having a 8-fold improvement over gold whereas in actual fact, the neuron was just further away. In addition to electrode performance characterisation, the proposed simultaneous optical/ electrical recording system has the potential to yield a wide range of data from neuronal populations such as extracellular spiking as well as VSD recorded neuronal parameters such as ion concentration membrane potential, subthreshold synaptic events and secondary messenger release [25]. The system could be further utilised to validate spike sorting algorithms through correlation of signal source and recorded microelectrode potential leading to the development of more robust detection methods. Conclusions A system to record from neurons both electrically, through multielectrode arrays, and optically via VSDs was developed. This system consisted of (i) MEA slides which were modified with PEDOT/PSS to reduce impedance and improve transduction properties, (ii) a custom built amplfication system which was capable of amplifying neuronal potentials with low intrinsic noise and (iii) an optical recording system which successfully visualised neuronal membrane potential changes through the use of a high speed camera and VSDs. Limitations of the described system are present within the microelectrode array used, the specifications of the PLOS ONE high speed camera and the use of an older generation of VSD, Di-4-ANEPPS. Utilisation of a MEA with microelectrodes separated by 50 μm would allow for the visualisation of many electrodes within one frame, improving the statistical power of the analysis. Furthermore, high speed cameras with faster exposure times (<1 ms) and a new generation of VSD would both facilitate the temporal resolution and image quality of the method. Electrically recorded action potentials were correlated with optical images and neuronal origin was confirmed via TTX. TTX addition also demonstrated the ability of optical imaging techniques to visualise subthreshold neuronal activity. The application of this system to microelectrode characterisation has highlighted discrepancies in recording performance when compared to traditional SNR calculation methods. The presented method allows for an incorporation of signal distance from the microelectrode tips, making quantification of recording performance more reliable. This system will allow for more in-depth studies on in-vitro neuronal populations through MEAs and will pave the path for validation of electrode performance and spike sorting techniques.	2020-08-22T13:01:28.465Z	2020-08-20T00:00:00.000	{ "year": 2020, "sha1": "f1f0f7e18aff6e717a330bca325f7111db9fb3c6", "oa_license": "CCBY", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0237709&type=printable", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "e9fee8b4140705da59189382cca12d76c7df72db", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine", "Materials Science" ] }
233314211	pes2o/s2orc	v3-fos-license	The evolution of life cycle assessment in European policies over three decades Purpose Life cycle thinking (LCT) and life cycle assessment (LCA) are increasingly considered pivotal concept and method for supporting sustainable transitions. LCA plays a relevant role in decision support, for the ambition of a holistic coverage of environmental dimensions and for the identification of hotspots, possible trade-offs, and burden shifting among life cycle stages or impact categories. These features are also relevant when the decision support is needed in policy domain. With a focus on EU policies, the present study explores the evolution and implementation of life cycle concepts and approaches over three decades. Methods Adopting an historical perspective, a review of current European Union (EU) legal acts and communications explicitly mentioning LCT, LCA, life cycle costing (LCC), and environmental footprint (the European Product and Organisation Environmental Footprint PEF/OEF) is performed, considering the timeframe from 1990 to 2020. The documents are categorised by year and according to their types (e.g. regulations, directives, communications) and based on the covered sectors (e.g. waste, energy, buildings). Documents for which life cycle concepts and approaches had a crucial role are identified, and a shortlist of these legal acts and communications is derived. Results and discussion Over the years, LCT and life cycle approaches have been increasingly mentioned in policy. From the Ecolabel Regulation of 1992, to the Green Deal in 2019, life cycle considerations are of particular interest in the EU. The present work analysed a total of 159 policies and 167 communications. While in some sectors (e.g. products, vehicles, and waste) life cycle concepts and approaches have been adopted with higher levels of prescriptiveness, implementation in other sectors (e.g. food and agriculture) is only at a preliminary stage. Moreover, life cycle (especially LCT) is frequently addressed and cited only as a general concept and in a rather generic manner. Additionally, more stringent and rigorous methods (LCA, PEF/OEF) are commonly cited only in view of future policy developments, even if a more mature interest in lifecycle is evident in recent policies. Conclusion The EU has been a frontrunner in the implementation of LCT/LCA in policies. However, despite a growing trend in this implementation, the development of new stringent and mandatory requirements related to life cycle is still relatively limited. In fact, there are still issues to be solved in the interface between science and policy making (such as verification and market surveillance) to ensure a wider implementation of LCT and LCA. Introduction The transition towards sustainability poses huge challenges to policy making, especially for what concerns comprehensive assessment of impacts and trade-offs between environmental pressures and impacts and measures to limit them. Policy makers are increasingly confronted with multifaceted challenges, in which environmental and socio-economic issues should be considered simultaneously and in which Communicated by Matthias Finkbeiner. the threats posed by the economic crises should be turned into positive actions towards sustainability. Life cycle thinking (LCT) is a systemic and comprehensive concept considered pivotal to provide support in better integrating sustainability into policy making (Pennington et al. 2007;Sonnemann et al. 2018). LCT is the basic concept referring to the needs of assessing burdens of products/ sectors/projects adopting a holistic perspective, from raw material extraction to end of life. LCT aims at avoiding the shift of burdens between environmental impact categories (e.g. reducing climate change while increasing land use), shifting between world regions (e.g. reducing domestic impacts while increasing spill-over effects), or between life cycle stages (e.g. decreasing impacts during the manufacturing of a product while increasing the impacts due to the end of life management) (Sala 2019). LCT is at the core of the Sustainable Development Goal (SDG) 12 of the UN Agenda 2030, "responsible consumption and production" that aims at the adoption of more sustainable patterns of consumption by the year 2030 and whose achievement needs a significant focus on the supply chain from producers to final consumers (UN 2015). According to the definition reported in the ISO standard (ISO 2006a), life cycle assessment (LCA) is the compilation and evaluation of the inputs, outputs, and the potential environmental impacts of a product system throughout its life cycle. LCA builds on a life cycle thinking perspective to provide valuable comprehensive quantitative information on the environmental performance of goods and services, and it can be effective to assess and support sustainable production and consumption (SCP), in particular as scientific basis for policies on products design, consumer information, public procurement, waste management, energy, and food supply (Sonnemann et al. 2018). However, a number of barriers still limit the use of LCA in policies, as the lack of widespread technical knowledge on LCA, and the lack of trust in LCA process and results (Seidel 2016). Moreover, to be enforceable in the context of entry market instruments, life cycle-based requirements need to also be verifiable by market surveillance authorities, e.g. through standardised methods (Mathieux et al. 2020). Flaws and variability in LCA studies, in particular in the interpretation phase, may both mislead and reduce the trust of policy makers in LCA (Agostini et al, 2020). Together with LCT and LCA, life cycle costing (LCC) is being applied by an increasing number of public authorities across the EU and in a range of sectors (EU, 2020a). LCC is defined as an "economic assessment considering all agreed projected significant and relevant cost flows over a period of analysis expressed in monetary value. The projected costs are those needed to achieve defined levels of performance, including reliability, safety and availability" (ISO 2017). The role of LCC is especially relevant in the context of sustainable public procurement, for demonstrating the best value for money and for governments, since the integration of LCC into policies is emerging as a relevant approach for comparing "green" and socially preferable assets with their less sustainable substitutes (Perera et al. 2009). Although facing key challenges, LCT and LCA have been increasingly implemented in policies around the world in the last three decades (Sonnemann et al. 2018). Over time some studies were developed to assess the level of implementation of LCA in policies (e.g. Curran 1997, Sonneman et al. 2018, in some cases focusing on specific cases (e.g. for EU, Sala et al. 2016, for USA, Schenck et al. 2009, or Japan, Hunkeler et al. 1998. Other studies focused on the specific needs of the LCA for policies, namely whether these needs are different from LCA used for assessment of product performance (Waardenaar et al. 2012), including the appropriateness of modelling approaches (e.g. whether consequential or attributional fit better the purposes as in Vázquez-Rowe et al. (2014) or Brander et al. (2008. Moreover, the role of LCA in sectorial policies was assessed, e.g. for waste as in Lazarevic et al. (2012) or in development policies as in Kituyi (2004). As of today, a number of regulations using LCT and LCA are implemented in several distinct regions of the globe. LCA has been applied in policy development and implementation in several countries, including the USA (through the Energy Independence and Security Act of 2007 (EPA 2020); Reed 2012), China (e.g. promoting the use of LCA in product design; Sonnemann et al. 2018), and Thailand (e.g. implementing LCA to assess the Thai Green Procurement Plan; Sonnemann et al. 2018). Latin America and the Caribbean region have developed its SCP strategy since 2003 and are considered a pioneer on SCP by the United Nations Environment Programme (UNEP 2015). Mexico, Chile, Colombia, and Brazil have used LCA in several policies, in a context of emerging awareness of policy-makers in these emerging economies (Valdivia et al. 2017;Maia de Souza et al. 2017;Güereca et al. 2015). However, the EU has been in the forefront in using LCA into development and application to a much larger extent than in any other region in the world (Sonnemann et al. 2018;Reed 2012). Current European initiatives on building life cycle-based policies are observed with attention from other parts of the globe, as stated, e.g. by Maia de Souza et al. (2017) for Brazil. In June 2001, the European Commission (EC) published the communication on Integrated Product Policy (IPP) (CEC 2001). This communication had a crucial role in paving the path for the development of policies including life cycle concepts, recognising life cycle thinking as an element contributing to sustainable development and science-based decision making. In December 2005, life cycle thinking exhibited a pivotal role as an indicator to monitor decoupling economic growth from environmental impacts in the Thematic Strategy on the Sustainable Use of Natural Resources (CEC 2005a). In parallel, the Thematic Strategy on the Prevention and Recycling of Waste (CEC 2005b) proposed an amendment (CEC 2005c) of the 1975 Waste Framework Directive (EC 1975) in view of introducing life cycle thinking for assessing environmental impacts of the generation and management of waste. These communications and legal acts were the basis for the implementation of life cycle concepts in European policies in the year 2000s. In particular, the Ecolabel Regulation (EC 1992(EC , 2010a) and the Ecodesign Directive (EC 2005(EC , 2009a constitute relevant LCA application in policies: the former aiming at establishing a voluntary ecolabel award scheme to promote products with reduced life cycle environmental impacts and the latter aiming at establishing consistent EU-wide rules for improving the environmental performance of energy-related products through ecodesign. The Ecodesign Directive was also a first worldwide example of the introduction of mandatory requirements for new products following on a life cycle approach, including the possibility of setting generic requirements based on the so-called "ecological profile." 1 A milestone in the development of policies based on LCA was the 2013 communication "Building the Single Market for Green Products" (CEC 2013a) and the linked Recommendation (EC 2013a), establishing the Product-and Organisation-Environmental Footprint methods (respectively, PEF and OEF). These LCA-based methods aim at the quantification of the environmental impacts of products and organisations, improving replicability, robustness and transparency of LCA studies. In fact, the PEF/OEF methods introduced an important improvement and guidance compared to the existing LCA standards (ISO 2006a, b) concerning key methodological choices and data quality requirements, and overall contributed to the comparability and verifiability of green claims by companies. Recently, the concept of life cycle has been at the centre of relevant policies (in sectors such as packaging and packaging waste (EC 2018a), vehicles (EC 2019a), and plastic (EC 2019b)) and communications (as for instance the European Green Deal (CEC 2019a), the Circular Economy Action Plan (CEC 2020a), and the Farm to Fork Strategy (CEC 2020b)), underlining the growing interest on life cycle thinking and life cycle approaches from the EU. In this wide policy context, it is of the utmost relevance to further improve the science-to-policy interface, due to the broad implications of the decisions supported by LCA. The importance of a stricter and more comprehensive implementation of LCA in policies is key to achieve effective benefits in a life cycle perspective and to avoid or limit burden shifting. A number of theoretical options for the implementation of LCA into policy already exists, although a consensus on proper technical requirements is missing (Lehmann et al. 2015). With this regard, it is worth mentioning in particular the role of the European Commission-Joint Research Centre (EC-JRC), which has been working towards providing this policy support through a number of projects and initiatives since 2004. In particular, the European Platform on LCA (EPLCA) (EU, 2020b) constitutes a key tool with this regard, having the objective to promote life cycle thinking in business and in policy making in the EU, as well as providing crucial support for the development of the PEF and the OEF (EU 2020c), in terms of methods and data. Based on the learnings of the EF pilot phase, organised by the EC in the period 2013-2018, the JRC published reports with suggestions for updating the PEF, including the methods for impact assessment. Methodological guidance and updates thereof could be found in the EPLCA (EU 2020c). This study aims at shedding light on the evolution and the relevance of life cycle thinking and life cycle approaches in policies, illustrating the trend in the use of these concepts and methods in the policy context of EU. This review aims at deepening the understanding on the link between policy and science, shedding light on shortcomings and possible future improvements, in view of supporting the evolution of policies from the adoption of general life cycle-related concepts to more comprehensive uses of LCA, and PEF/OEF methods. This review systematically explores the historical development of the implementation of LCT, LCA, LCC, and environmental footprint and the way they have been integrated in the EU policy context in strategic and legal acts, analysing the period ranging from 1990 to October 2020. Methods A review of EU policy documents was performed in this study as a basis to assess the evolution of occurrence of LCT concepts and LCA-related approaches over time (including also LCC and PEF/OEF). The study focused on the EU due to the wide development of policies mentioning life cycle approaches and methods and considering the presence of a comprehensive online database for the collection of the documents (see Sect. 2.1). To this end, a search customised to a number of different EU policy documents has been conducted. EU treaties are achieved by several types of legal acts that have a different level of prescriptiveness, a different set of requirements and involvement of the Member States (MS) (EU 2020d, e, f, g; EUR-Lex 2020a). This work focused on EU "policies" and on "communications." In this study, documents such as proposals, reports, white papers, and green papers are included under the label "communication". The complete list of definitions of legal acts included under the "policies" and the definition of "communications," are reported in the Supplementary Information (SI). Communications were kept separate from the legal acts considering that these documents do not constitute legally binding acts but are rather strategic documents "suggesting" actions or "proposing" legal acts. Standards were not included in the analysis given that the scope of the work was dedicated to legal acts (both legally binding and non-legally binding). Research method This review was based on a search performed through EUR-Lex (EUR-Lex 2020b), the online database that provides access to all EU policies (regulations, directives, decisions, etc.) and communications to identify EU policies referring to LCT, LCA, and LCC. Given its more limited development and application in policies, social LCA was not included in this analysis. A number of keywords related to the concepts of LCT and LCA have been screened within title and text of both EU policies and EU communications (e.g. "life cycle thinking," "life cycle assessment"; the details on the search string are reported in the SI), and EU policies or communications presenting these keywords were collected for the period ranging from 1990 to October 2020. Furthermore, other policies and communications identified as potentially relevant (not found through the abovementioned screening but cited in other documents or identified by the authors of this article through their expert judgment) were added to the overall set of legal acts to be investigated. A refined selection was then performed on collected documents, aiming at excluding documents out of the scope of the present review (e.g. documents dealing with "life cycle" of insects, plants, or other organisms/species; the "life cycle" of projects/programmes, referring to the "life cycle" of people and workers): this refinement was manually performed, critically analysing the context of the legal documents previously gathered. Classification method Policies and communications were classified according to the concept and/or method cited in the document, focusing on (i) LCT (life cycle thinking), 2 (ii) LCA (life cycle assessment), (iii) LCC (life cycle costing), and the European Product and Organisation Environmental Footprint (PEF/ OEF). The presence of more than one concept/method was considered (e.g. LCT and LCA, both mentioned): in these cases, the document was classified under multiple concepts and/or methods (for instance the new Circular Economy Action Plan (CEC 2020a) was classified under LCT, LCA, and PEF/OEF). The classification was structured according to different perspectives: (i) the relevance of life cycle concepts in policies and communications, (ii) the sectors addressed in policies and communications, (iii) the presence of links to case studies and previous legal acts in communications, and (iv) the specific types of applications of life cycle approaches. Regarding the relevance of life cycle concepts in policies and communications, in view of understanding the importance and prominence given to LCT, LCA, LCC, and/or PEF/OEF in the analysed policies, a further classification was performed: 1. LCT, LCA, LCC, and/or PEF/OEF cited "at the heart" of the document: if the concept/method exhibited a pivotal role and/or was a crucial element for the document itself (e.g. not only mentioned in the preamble, but instead used to set a requirement, an objective for future developments, a selected criterion to be adopted when assessing environmental impacts) or if it was mentioned with a key role in the context of future additional legal acts to be developed; 2. LCT, LCA, LCC, and/or PEF/OEF cited "in the context" of the document: in all the other cases where, for instance, the concept/method was cited in the preamble (e.g. in the recitals) and in other sections in a generic manner, or if the policy was only referring to these concepts as cited into other existing documents. From the complete list of policies and communications, a specific subset of key documents was selected in view of providing a dedicated and detailed overview of the most relevant policies and communications where LCT, LCA, LCC, and/or PEF/OEF exhibit a crucial role. Furthermore, we performed an analysis of requirements in policies based on a life cycle approach (e.g. requirements related to LCT, LCA, LCC, PEF/OEF). Additional details on the analysis of these requirements are included in tables S4 and S5 of the SI. Concerning the sectors addressed in policies and communication, all legal acts were classified focusing on their specific sectors of concern (e.g. waste, energy and fuels, construction, etc.). Considering the total number of policies and communication addressed in this review, the most frequently mentioned sectors and field of action (e.g. waste, energy and fuels, construction) were identified and commented. Whenever possible, the same categories were considered for the classification of communications and policies, in order to ease the comparison between these two types of documents (highlighting both similarities and differences). Specific sectors mentioned several times only in policies or only in communications were also included. Moreover, general categories (e.g. "others not elsewhere classified (n.e.c.)") were established for policies addressing specific products and specific sectors not represented in multiple documents, or policies related to products in general (e.g. integrated product policies). Similarly, a generic category was also established for communications related to research and other specific programmes (such as funding programmes, e.g. "Horizon 2020"). In the classification procedure, in case of the presence of more than one sector for a single legal document, the latter has been classified in several categories (e.g. a classification within both the "waste" and "energy and fuels" categories). Further details on the abovementioned classifications of policies and communications are reported in the SI. If a single document is relevant for more than one of the selected sectors/identified scope, it is classified accordingly with multiple entries (e.g. the Directive on the promotion of the use of biofuels or other renewable fuels for transport (EC 2003) was classified in the sectors "energy and fuels" and "mobility"). Furthermore, policies were also investigated in view of identifying any relevant case in which LCA, LCC, and/or PEF/OEF were directly applied in the policies or indicated as the reference method for conducting an assessment. On the other hand, communications were further analysed to investigate the links with existing or foreseen legal acts. In case of recent proposals, not always the proposed policy has already been set up. In some of these cases, the proposals were identified as "possibly relevant in future," whenever the proposed policy was identifiable as possibly relevant with regard to future development/implications of the LCT concept or of the LCA, LCC, PEF/OEF methods. The communications were also categorised in view of providing insights on the specific type of application of the life cycle approach, to whether they were referring 1. to LCA, LCC, and/or PEF/OEF results: when results from life cycle analysis (i.e.: LCA, LCC, PEF/OEF) are mentioned in the analysed document or are cited from other studies; 2. to LCT, LCA, LCC, and/or PEF/OEF as implemented in other legal acts, currently or in the past: when the role of life cycle thinking and life cycle approaches are mentioned as being pivotal in other legal acts or when other legal acts -where these concepts are relevant-are cited; 3. to LCT, LCA, LCC, and/or PEF/OEF as important to be implemented, in the future: when it is explicitly mentioned that life cycle thinking and life cycle approaches are crucial elements to be considered and implemented in the future. A parallel classification was conducted by excluding specific subsequent legal acts (as well as other linked documents, such as working plans or amendments) derived from the implementations of the Ecodesign Directive (EC 2009a) and the Ecolabel Regulation (EC 2010a). 3 The scope of this analysis was to unveil individual initiatives removing those that were implementations of existing ones. The results of this analysis are presented in the SI. Results The following sections illustrate the results of this study, following the methodological approach presented in Sects. 2.1 and 2.2. Results of the search According to the method described in Sect. 2.1, overall, a total of 243 policies and a total of 348 communications were gathered through the EUR-Lex search, the screening of the citations inside each reviewed document, as well as those identified via expert judgment. After this first step, documents not relevant for the present review were discarded, resulting in a total of 159 policies and a total of 167 communications that were further analysed and classified as presented in the following sections. Results of the classification In order to better understand the importance given to the life cycle related approaches cited in the selected documents, a distinction between the life cycle concepts/methods cited "at the heart" or "in the context" of the legal acts classified was implemented (see Sect. 2.2). Overall, LCT/LCC/LCA and/or PEF/OEF were found to be "at the heart" in 60% of the policies classified and in 45% of the communications classified (see Sect. 2.2). Table 1 presents an overview of the most relevant policies and communications where LCT, LCA, LCC, and/or PEF/ OEF are crucial, starting from the year 2000. A detailed description of the policies and communications included in Table 1 is presented in the SI, whilst additional comments are reported in the Discussion section. In particular, additional details on the contents of the most relevant classified policies and communication listed in Table 1 are reported in the Table S4 in the SI. In Table S4 each relevant policy and communication is listed from the oldest to the newest. Information regarding the linked policies/communications (and amendments, etc., if any) together with the assessed life cycle concepts/approach are specified. A brief description of the policy/communication itself is also reported. the results of the classification when removing these legal acts from the overall picture (the results of this analysis are presented in the SI). Decisions and regulations cover the majority of the policies classified, followed by directives (in line with general EU legal acts statistics, EUR-Lex 2020c), while Recommendations represent only a very minor share. Figure 2 highlights the results of the classification of sectors of intervention of the identified policies. The specific description of the sectorial categories considered in Figs. 2 and 3 is reported in the SI. Household goods (e.g. hygiene products) and appliances (e.g. vacuum cleaners, lamps) manifest an evident dominating role, mainly as specific implementations of the Ecolabel Regulation (in case of Decisions) and the Ecodesign Directive (in case of regulations). However, it is worth noticing that the dominance of this category was also due to its intrinsic heterogeneity, since it covers a wide array of different products. Moreover, the "research and other programmes" sector is particularly relevant (e.g. funding programmes such as "LIFE" and "Horizon2020" and other specific EU programmes, such as the European statistical programme), followed by "energy and fuels" and waste. Figure 3 reports the results of the classification for the sectors as conducted on the EU communications. In fact, communications show a coverage of a wide array of different sectors and topics , for which it was not always possible to identify a specific category. However, when a single sector category was identifiable, a predominance of communications dedicated to "energy and fuels" and "circular economy and sustainable development" is evident. Although communication has a broader scope than the classified policies, energy was found to be a key and transversally relevant sector in both policies and communications (Figs. 2 and 3). Only in a few cases, LCA, LCC, and/or PEF/OEF have been directly applied in the policies or indicated as the reference method for conducting an assessment. Decisions The key role of LCA has been further acknowledged in the development of the PEF/OEF methods, which have been notably defined in the 2013 EU Recommendation (EC 2013a) and which increasingly represent the reference LCA methods for the EU policies. LCC was instead considered as the reference method in policies dealing with public procurement (e.g. directives: EC 2014b, 2014c) and for the definition of Ecodesign requirements (e.g. the 2009 Directive EC, 2009a, for energy-related products). However, LCC was not described in detail in these legal acts: life cycle costs were mentioned as a relevant aspect to be monitored and minimised, whilst specific details on the advantages and disadvantages and information on the implementation of a life cycle costing method were neglected. Only 27 out of 167 classified communications referred to LCA, LCC, and/or PEF/OEF results. Existing LCA studies were cited as general examples (e.g. CEC 2008a) regarding the cost effectiveness of green public procurement in sectors where green products are not more expensive than nongreen alternatives. The communication (CEC, 2014) cited sectorial studies on the environmental benefits of recycling (e.g. gypsum board production), especially with regard to the achievable reductions of the global warming potential. In addition, LCA, LCC, and/or PEF/OEF results were also mentioned in the context of greenhouses gas emission savings (e.g. communications: CEC 2010a, 2010b). In 80 communications it was identified a specific reference to the relevance of implementing a life cycle approach. A number of communications (namely CEC 2013a, 2015a, 2018d, 2018b) cited the PEF/OEF methods, and in particular, the 2013 communication (CEC 2013a) on building the Single Market for Green Products which set the framework for the PEF/OEF methods to calculate robust and comparable information on the environmental performance of products and organisations. A dedicated analysis was additionally performed (Table S5, S1) to assess the level of implementation of life cycle-based requirements in policies. A total of 29 legal acts were analysed in detail. It resulted that, in general, strict requirements for life cycle methods are missing. Even when LCA or LCC are required or strongly suggested, precise guidelines for their implementation are not commonly provided. For instance, in the case of financial rules applicable to the general budget of the EU (contracts awarding; EC 2015b, c), a list of elements to be covered by a LCC study were indicated, but specific methodological requirements on how to perform the study were not provided. Furthermore, the directive (EC 2019b) defined measures to reduce life cycle environmental impacts of single-use plastics, although no further detailed information was reported on how to perform these assessments. On the other hand, a strong preponderance of LCA and LCC in policies related to environmental practises was noticeable, with a number of legal document (e.g. EC 2017b, 2018c, 2019e) that prescribed using LCA/LCC studies as benchmarks of excellence. In some cases (e.g. EC 2019f), the ISO standards (ISO 2006a, b) were mentioned as the reference guidelines according to which a LCA study should be performed. Lastly, PEF/OEF were extensively described in EC (2013a), illustrating in the annexes the complete methodology for performing PEF/OEF-compliant studies. The EC (2013a) was also one of the few cases in which a comprehensive LCA-based method was fully described in a legal document. Among the 167 classified communications, 25 were proposals (for a regulation, for a directive, for a decision) resulting in other legal acts already finalised and published. 4 On the other hand, other proposals related to policies envisioned in the future but not already operating were analysed. In the latter case, it is worth mentioning the 2020 communication (CEC 2020e), that demanded the active engagement of stakeholders at all levels of governance, to ensure that EU climate and environment laws are effectively implemented. This communication forms the EU's basis for achieving the United Nation's 2030 Agenda and its Sustainable Development Goals (UN 2015). Lastly, 2 communications, also explicitly referred to life cycle-related potential trade-offs. Notably, in the 2011 flagship initiative for a resource-efficient Europe (CEC 2011a), it was mentioned that in order to make the right choices with regard to resource-efficiency policies, there is the need to consider the whole life cycle of the way resources are used, including along the value chain, and the trade-offs between different priorities. Moreover, in the 2005 Thematic Strategy on the prevention and recycling of waste (CEC 2005b), it was stated that all phases in the life cycle of a resource need to be considered as there can be trade-offs between different phases, and measures adopted to reduce environmental impact in one phase can increase the impact in another. Despite being a crucial aspect to be considered and understood when interpreting life cycle-based comparative results (Prado-Lopez et al. 2016), the results of our analysis on EU communications show that trade-offs are generally implicitly taken into consideration when referring to life cycle approaches. Discussion Although the classification and interpretation of the results were affected by a certain level of subjectivity, the collected information are nonetheless useful for providing qualitative insights on the embodiment of life cycle thinking and life cycle approaches in EU policies and communications. Moreover, these results could be used to provide recommendations for future improvements. The EU has already made significant steps forwards in the implementation of life cycle in policies, through various documents starting with the Integrated Product Policy (IPP) (CEC 2001(CEC , 2003, as it is possible to notice from Table 1 and the S1. In the abovementioned IPP, the EC concluded that LCA provides the best framework for assessing the potential environmental impacts of products that is currently available. However, the need for more consistent data and consensus on LCA methodologies was underlined. Later, LCT and LCA were then further integrated in specific sectorial policies, as the Thematic Strategy on the Prevention and Recycling of Waste (CEC 2005b), the Thematic Strategy on the Sustainable Use of Natural Resources (CEC 2005a), the Sustainable Consumption and Production, and Sustainable Industrial Policy Action Plan of 2008 (CEC 2008b). In 2005, the IPP communication was particularly strengthened by the Thematic Strategy on the Sustainable Use of Natural Resources of the EC (CEC 2005a), which focuses on decoupling economic growth from environmental impacts. LCT is core to this thematic strategy, being a foundation of the indicators that will be developed to monitor progress across the community. In the first decade of the 2000s, LCT was mentioned as a key approach in the Sustainable Consumption and Production and Sustainable Industrial Policy (SCP/SIP) Action Plan (CEC 2008b), as well as more overarching policy documents, such as the Resource Efficiency Flagship Initiative of the Europe 2020 Strategy (CEC 2011a), and related Roadmap (CEC 2011b). Of particular note here is the stated objective of the Roadmap to a Resource Efficient Europe according to which, by 2050, the EU economy shall have developed in way able to accommodate resource constraints and planetary boundaries. In 2013, together with the Recommendation establishing the PEF/OEF methods (EC 2013a), a landmark communication was released: the Single Market for Green Products communication (CEC 2013a). Among all others, the PEF/OEF methods constitute one of the most relevant game changers in the field of LCA. In fact, these are the first systematic attempts to have a robust and comparable method for LCA applied in policies and green claims, even from a legal point of view. The PEF and OEF methodologies build on existing LCA-based methods aiming at harmonising them and at establishing a tool to assess the environmental performance of products, services and companies based on a comprehensive assessment of environmental impacts over the life cycle. The 2013 EC Recommendation promoted the PEF/OEF in view of measuring and communicating the life cycle environmental performance of products or organisations. The final goal was to reduce and overcome the fragmentation of available methods for measuring environmental performances. In view of enabling comparability of products and organisations performances, the Product Environmental Footprint Category Rules (PEFCRs) and Organisation Environmental Footprint Sector Rules (OEFSRs) were developed by a Technical Secretariat, composed of at least 51% of the EU market, plus other stakeholders. Currently, 19 PEFCRs (covering for instance food, batteries, laundry detergents, metal sheets, laundry detergents, etc.) and 2 OEFSRs (related to copper production and retail) are available (EU 2020h). The development of a PEFCR/OEFSR is a crucial aspect for improving the robustness of the PEF/OEF, since these rules are established on the basis of an agreement between the scientific community and the industry. However, even if more PEFCRs are expected to be developed in the next years, the establishment of a PEFCR is a challenging process. The development process is commonly time-consuming and the final agreement between the involved parties on a specific PEFCR is not ensured, therefore limiting the potential prompt implementation in policies. In the most recent years (from 2016 onwards, with particular emphasis in the years 2018, 2019, and 2020), the number of policies and communications that are explicitly citing concepts and methods related with life cycle is growing. It is definitely worth of mention the 2019 communication (CEC 2019a) on the European Green Deal. This communication set the commitment of the EC to tackling climate and environmental-related challenges, aiming to transform the EU into a fair and prosperous society with a modern and resource-efficient economy and setting the target for the carbon neutrality of the EU by 2050. A number of other communications are also going in the same direction, complementary to the Green Deal by tackling some specific topics, as illustrated in Fig. 6. For instance, in case of the food sector (CEC 2020c), the relevance of a life cycle viewpoint was outlined, since a better information from "farm to fork" is foreseen (covering where the food comes, its nutritional value, and its Fig. 6 An overview of the role of LCT, LCA, and PEF/OEF within the EU Green Deal and related policies initiatives environmental footprint). Chemicals (CEC 2020d) were furthermore considered, with the aim of ensuring sustainability by minimising their environmental footprint. In the case of industry (CEC 2020f), a reduction of their carbon footprint ensuring sustainability was also envisaged. Moreover, circular economy was mentioned in view of proposing sustainable solutions to reduce EU consumption footprint, as it is detailed in the related communication (CEC 2020a). The measures proposed by the EU Green Deal will significantly benefit EU companies, in particular by means of and improved digitalisation and information flow, providing more detailed information on product's origin, composition, repair and end-of-life handling, therefore enabling a more circular economy. This will improve the reliability of companies' "green claims," therefore reducing the risk of "greenwashing" and the proliferation of unreliable and not scientifically based environmental labels that are leveraging on the worldwide growing interest on sustainable topic. Lastly, many other fields of action (e.g. product policies, investments, biodiversity) were addressed as well in the European Green Deal, identifying this communication as a key document for the future of Europe. The analysis of the abovementioned documents indicated that no case study or benchmarks are mentioned and that they do not include any specific detail on how to perform a LCA. However, despite not proposing specific requirements, a stronger commitment to the implementation of LCA-related results is evident. From Table 1, the main key sectors of the classified policies and communications can be identified. In particular, both the waste sector (e.g. communication: CEC 2005b; directives: EC 2008, 2018a) and the construction sector (e.g. regulation: EC 2011; communication: CEC 2012a) can be highlighted, since in these cases, life cycle approaches and methods, in particular LCA and LCC, exhibit particular relevance. Most of the abovementioned policies and communications focused mainly on LCA, whilst LCC was especially mentioned in specific context, such as the directives on procurement (EC 2014a, b, c) or on specific sectors, e.g. the construction sector. For instance, in the 2012 strategy for the sustainable competitiveness of the construction sector and its enterprises (CEC 2012a), LCC was mentioned as a methodology for supporting the development of a model for Green Public Procurement and for sustainable development principles related to regional policies. In two other fields, Green Public Procurement and Sustainable Public Procurement (communication: CEC 2008a; directives: EC 2014a, b, c), life cycle concepts have pivotal roles in the procurement processes themselves. Moreover, in the 2017 proposal for a regulation on low carbon benchmarks and positive carbon impact benchmarks (CEC 2018b), that resulted in the 2019 regulation (EC 2019g), the PEF/ OEF methods were remarkably envisioned as "the preferred option" among the future policy alternatives for harmonising the methodology to be applied to low-carbon indices and "positive carbon impact" indices to help investors compare the carbon footprint of investments. In addition, other relevant sectors in the field of life cycle were identified. This is particularly the case of energyrelated policies: the 2005 Directive (EC, 2005) on Ecodesign requirements for energy-using products, the 2016 communication (CEC 2016a) on accelerating clean energy innovation, the 2017 regulation (EC 2017c) on energy labelling, the 2018 RED II Directive (EC 2018b) on renewable energy, and the 2019 Directive (EC, 2019a) on the promotion of clean and energy-efficient road transport vehicles and communications specific to batteries (e.g. communications: CEC 2018e, 2019b). The analysis of life cycle concepts and approaches in the policy and communication documents for the energy category highlights how the main focus is related to the transition towards the use of the renewable energies and fuels. The directive of 1998 relating to the quality of petrol and diesel fuels (EC 1998) has been updated and mentioned in recent years (EC 2009c(EC , 2015a, especially in the field of the monitoring and the reduction of life cycle greenhouse gases emissions and the improvement of the quality of petrol and diesel fuels towards the promotion of the use of renewable energy sources. With this regard, the promotion of biofuels and other renewable fuels was proposed already in 2003 (EC 2003) introducing minimum proportion of biofuels and other renewable fuels placed on the Member States markets. Moreover, transversal issues related to the adoption of biofuels were mentioned, such as the indirect land-use change related to biofuels and bioliquids as indicated in the report of 2010 (CEC 2010a). Lastly, recent document stemming from the European Green Deal (CEC 2019a) focused on alternative energy sources, such as hydrogen (CEC 2020g), which is considered a key energy carrier for the reduction of life cycle greenhouse gas emissions by minimum 50% and towards 55% by 2030, in a cost effective way. Furthermore, two policies have been also relevant concerning green purchasing: (i) the 2016 communication (CEC 2016b) that resulted in the 2018 regulation (EC 2018e) on the Governance of the General Union, which cites that MS shall require a life cycle thinking approach to energy suppliers in order to reduce life cycle greenhouse emissions, and (ii) the 2017 communication (CEC 2017a) related to the EU Environmental Implementation Review, where a LCT perspective was presented as a successful example of an approach (that has been adopted by many MS) able to make purchased goods and services more sustainable through public procurement. Whilst Table 1 provides insights on the main key legal acts analysed, Fig. 1 illustrates that decisions and regulations were essential legal acts with regard to the entirety of the classified policies (a total of 159). Decisions frequently were related to the "research and other programmes" sector (for instance related to the adoption of the multiannual LIFE work programme or the establishment of the specific programme implementing "Horizon 2020") and were also commonly present with regard to the household goods sector (given the numerous specific applications of the Ecolabel Regulation) (Fig. 2). Regulations on the other hand are the most binding EU policies acts (see the SI): LCT/LCA/LCC have been crucial into regulating appliances (in particular the numerous specific applications of the Ecodesign Directive) as well as in the "research and other programmes" sector (for instance establishing the "Horizon2020" programme programmes or concerning LIFE + programme) (Fig. 2). With regard to directives, LCT/LCA/LCC have been crucial for the waste sector (e.g. waste, packaging and packaging waste, Waste Electrical and Electronic Equipment-WEEE) and for the "energy and fuels" sector (e.g. petrol and diesel fuels, biofuels, and other renewable fuels for transport). However, compared to regulations, the classification of EU directives showed wider application of LCA to different sectors (e.g. waste, mobility, energy, and fuels) (Fig. 2). It is however important to keep into consideration that beside their frequency, the content of the classified legal acts was relevant as well. For instance, results indicate that Recommendations were less represented compared to other legal acts. However, for instance the Recommendation of 2013 (EC 2013a) introduced the PEF/OEF methods, which had an outstanding relevance in the following LCA-related policies and communications, especially in the more recent years. Most of the classified communications (Fig. 3) have been mainly implemented with regard to (i) "energy and fuels" (e.g. focusing on clean energy innovation and a rational use of energy, and fuels such as biofuels and hydrogen) and (ii) "circular economy and sustainable development," mainly in recent years (due to the increasing focus on these topics, noticeable for instance in the European Green Deal in 2019, in the Circular Economy Action Plan and in the Chemicals Strategy for Sustainability in 2020, etc.). The idea of circular economy has roots in concepts of real-world systems and cycles, where waste of resources is minimised and materials are maintained for as long as possible "in circle," therefore reducing waste generation. In recent years, the EU has promoted the transition towards a circular economy, to achieve a regenerative growth model able to maintain production and consumption systems within sustainable levels. Most of the policies and communications discussing the concepts of Circular Economy were published in the years from 2019 to 2020, leveraging on two key earlier documents: the EU action plan for circular economy of 2015 (CEC 2015a) and the communication on the role of waste-to-energy in the circular economy of 2017 (CEC 2017b). Furthermore, waste (e.g. end-of-life vehicle, waste management) and "industry and business" (with a preponderance of documents related to public procurement) are two other relevant sectors among the analysed communications, together with "research and other specific programmes," that are assessing a wide array of topics (e.g. nanotechnology, the urban environment, economy in the EU Atlantic area). In general, results highlight that life cycle concepts (especially LCT) have been primarily developed and implemented in product policies, showing a clear higher level of maturity in the implementation in this field compared to others. Lastly, from Figs. 4 and 5, it was evident the presence of early legal acts including life cycle considerations. In fact, the first ISO standards formalising the LCA methodology are dated 1997-1998(ISO 1997, which were then substituted by the more recent standards (ISO 2006a, b). Results indicate the presence of policies and communications which already contained LCT concepts before 1997. Notably, a LCT approach was already mentioned in 1991 in the decision related to research and technological development in the field of industrial and materials technologies (EC 1991), in which considerations on the whole products life cycle were seen as crucial for a revitalisation of EU manufacturing processes. In the regulation of 1992 (EC 1992), the eco-label award scheme was established for the first time. This regulation focuses on the design, production, marketing, and use of products which have a reduced environmental impact during their entire life cycle. This regulation was anticipated by the related proposal in 1991 (CEC 1991). Furthermore, in 1994 the directive (EC 1994) on packaging and packaging waste was published, in which it was stated how the Commission shall promote the preparation of EU standards relating to criteria and methodologies for life cycle analysis of packaging. It is also worth noticing how the environmental footprint was already mentioned even when the method was not precisely defined yet (i.e.: before the 2013 recommendation, EC, 2013a). In fact, some earlier communications (CEC, 2012d(CEC, , e, 2013b highlighted how the Commission intended to recommend the environmental footprint as best practices and methodologies in the field of life cycle assessment. Furthermore, the application of LCA for macro-scale analysis has increasingly gained interest in the policy context. EC-JRC has developed a set of life cycle based indicators for assessing the environmental impact of consumption in Europe (Consumption and Consumer footprint, Sala et al. 2019). An example of the use of these indicators, is in the Circular Economy Action Pan (CEC 2020a) which mentions consumption footprint in the section on monitoring. However, also other policies refer to the need of reducing the overall footprint, i.e. in the case of the "farm to fork" communication (CEC 2020c), aiming at the reduction of the overall footprint of the food system. Even if LCA, LCC, and especially PEF/OEF are increasingly relevant, LCT is continuosly mentioned in policies, in more and more sectors of intervention. It is nonetheless worth mentioning that LCT exhibits an overall slight reduction in its yearly frequency when the 57 decisions related to Ecolabel criteria for specific products and the 24 regulations related to specific Ecodesign implementations are removed (see the SI). This aspect constitutes a particularly relevant element of reflection: in fact, in many cases, "life cycle" was addressed and cited only as a general and rather unspecified concept, while more stringent and rigorous methods (such as LCA and PEF/OEF) were commonly cited in view of future development or within few key regulations and/or directives (e.g. the EMAS regulation (EC 2009b), the renewable energy Directive (EC, 2018b), the "single use plastic" Directive (EC 2019b), the sustainable investments regulation (EC 2020a)). A number of key aspects can be identified to ensure applicability of LCA into policies. When life cycle assessment is applied to specific requirements (i.e. those set into products' policies), it is crucial to grant transparency, reproducibility and verifiability of the results. Based also on similar experiences on the inclusion of environmental requirements into policies (e.g. those related to raw materials sourcing, durability, end-of-life management (Ardente and Mathieux 2014)), a system for the verification of LCA results is probably the most relevant challenge to overcome. In fact, an important aspect worth considering is that in view of ensuring clear (mandatory) LC-based requirement in policies, these requirements should be both more binding but also verifiable. The method for the verification of LCA results in policy requirements (necessary, for example, to prove the achievement of a certain target) needs to be sufficiently robust, being potentially suitable "for suing for trial" a non-compliant company. In particular, to grant robustness and verifiability, it is necessary to guarantee availability and quality of LCI (life cycle inventory) data. Other methodological and practical issues can also affect the integration of LCA into policies, such as the following: 1. The use of a consequential versus attributional LCA (depending on the scope of the policy). So far, consequential LCA concepts have been implemented only to a limited extent. In particular, it is worth mentioning the Renewable Energy Directive REDII (EC 2018b) that accounts for indirect land use change; 2. The relationship and integration with other models widely used in policy context (e.g. economic models such as partial equilibrium models or general equilibrium models, or other qualitative/quantitative methods for environmental impact assessment); 3. The need of normalization and possibly weighting factors when using impacts assessment methods addressing different impact categories; 4. The need of defining benchmarks for evaluating policy options (especially in light of using LCA in some label-ling schemes, or to set impacts thresholds beyond which the product is not acceptable in the market). Despite the recent development of the PEF/OEF, policies and communications are frequently missing specific methodological and data requirements together with key detailed explanation on steps and procedures to be followed when performing a LCA study. The results of the dedicated requirement analysis (SI) indicated an overall lack of strict requirements or methodological guidelines for performing life cycle evaluations. This highlights how LCA should be better integrated in policy making, in a way that also ensures the right balance between the following: 1. Enhancing the comparability of LCA results (by being very prescriptive in data and methodological choices) versus providing the required flexibility (in order to grant the applicability of the LCA to diverse product groups or sectors); 2. Allowing limited assessments for a few impact categories (simplifying the interpretation of the results) versus pushing towards more comprehensive assessments (including a large set of impact categories but more difficult to be assessed jointly); 3. Using only very robust life cycle impact assessment methods (e.g. with regard to Global Warming Potential, as the main used in policies) versus extending to methods able to address more impacts (e.g. 16 environmental impacts as recommended by PEF), ensuring their applicability into policies. Conclusions Life cycle thinking and assessment are progressively moving from academic implementation and ad hoc uses (primarily in-house in large companies) to broader applications in the society. This article presents one of the first systematic review and assessment of the applications of the LCT/LCA/ LCC into policies, with a special focus to the EU context. Over the last 30 years, there has been an increasing emphasis on integrated approaches in environment policy in the EU. Policies have focused on linkages between environment media (such as air, water and soil) and cross-cutting environment themes (e.g. climate change, biodiversity) that pay more attention to sustainable resource use. In a growing number of policies, LCT and LCA have been recognised as useful approaches supporting impact assessment, implementation measures, and monitoring needs. Already in 1990, the Council resolution of 7 May 1990 on waste policy (EC 1990) invited the Commission to submit as soon as possible a proposal for a Community-wide ecolabelling scheme covering the environmental impact during the entire life cycle of the product. This resulted in the first EU Regulation regarding Ecolabel (CEC 1991;EC 1992), in which it was evident how the evaluation of the impacts associated to product life cycle is at the core of the label scheme. From the results of the present study it is evident the pivotal role not only of the Ecolabel Regulation (EC 1992(EC , 2010a, but also of the Ecodesign Directive (EC 2005(EC , 2009a. These two policies have been important in the widespread of LCA with regard to products, since a high number of specific documents implementing these policies have been introduced during the years. In addition, recent cutting-edge methodologies (the PEF/OEF) manifest the growing interest in the harmonisation of key methodological choices and data quality requirements in the field of LCA. In the most recent years (from 2016 onwards), the number of policies and communications that are explicitly citing concepts and methods related with life cycle is growing, and the EU policy-making can be considered a frontrunner in the implementation of LCA into policies. Based on the policies classified in this review, it is evident that the concept of life cycle has been continuously growing in number and relevance of the applications, and it is increasingly cited and envisioned as a relevant element also for future policies. In particular, LCT is very horizontally adopted in key strategic documents for setting of new policies and ambitious targets in the context of the EU Green Deal. However, it was identified that life cycle concepts could be implemented in a more thorough way in policies. In fact, results of a dedicated assessment of life cycle-based requirements in policies indicate a general lack of specific methodological requirements and guidelines for the application of life cycle methodologies (LCA, LCC and PEF/OEF). Depending on the type of application of policies, even the quality of LCA results can change. For general policies (including strategic document, planning, research programmes), it is still possible to use LCA to a higher level of uncertainty. When LCA is instead applied to specific requirements, it is necessary to specify with a very high level of details the methods and underlying data to be used. The presence of market surveillance authorities of MS could ensure that such requirements are binding enough, demanding for instance compulsory information. The abovementioned aspects are crucial for any policy makers and require actions by several stakeholders (from methods developer to policy maker) to ensure applicability of LCA to policies. Furthermore, a number of other methodological and practical issues that can affect the integration of LCA into policies can be identified (e.g. the use of consequential or attributional LCA, the need of defining benchmarks, the integration with other models widely used in policy context). An overall objective that is possible to identify from the performed analysis is the need of moving from general LCT approaches, which have been to date predominant compared to other life cycle approaches, toward more comprehensive uses of LCA (such as, for instance, evaluation of impacts according to the prescription of the Environmental Footprint methods). Among all others, the PEF/OEF methods are worth mentioning, since these represent one of the most relevant initiatives in the field of life cycle assessment in recent years. These methodologies seek to establish a common method to measure and communicate the life cycle environmental performance of products and organisations, and, furthermore, introduced the PEFCRs. Given the wide range of different decision-contexts and sectors, the development of product-/sector-specific criteria, guidelines, and simplified tools would provide an important support for fostering LCA applications, and the PEFCRs are going into this direction. In fact, PEFCRs will support practitioners in focusing on the most important parameters, thus also reducing the time, cost and the expertise required. In general, the main field of action where life cycle was found to be relevant in the EU set of policies referred to product policies. This is evident, for instance, in the relevance of the Ecolabel Regulation and the Ecodesign Directive (and related implementations), as illustrated in the results of this work. The importance of the evaluation of environmental impacts/costs with a life cycle mindset is expected to further grow and the analysis of recent policies and communication indicate that the environmental footprint methods will be strongly considered in the foreseeable future. Finally, there is the need of collaboration and co-production of knowledge towards harmonisation of life cycle approaches. This should be done in order to ensure that different policies considering life cycle perspectives are coordinated when assessing diverse topics. This implies also additional efforts towards harmonising life cycle methods adopted in different policy contexts which have a life cycle perspective. In this context it is worth mentioning the relevant role of the scientific community, which should keep in mind the policy needs when proposing for instance new impact assessment method. In fact, it would be important to account for robustness, data availability, representativeness, understanding (for non LCA-experts), and verifiability in developing new LC-related results, in view of pave the paths for easier policy implementations. Our expectation, based on the past experience of using LCT and LCA approaches for policy support, is that the use of life cycle methodologies and related methods and tools in policy support will continue to grow in influence in the foreseeable future and that the environmental performances of MS will be improved by the time of the next review. Acknowledgements Beyond the authors, a range of (existing and former) staff members of the European Commission Joint Research Centre have contributed to an increasing use of LCA in EU policies and in laying the foundations of this this work, namely David Pennington, Rana Pant and Fabrice Mathieux. We also acknowledge the support of the Directorate General Environment of the European Commission for partly financing the described activities via the Administrative Arrangements Technical support for the Environmental Footprint and the Life Cycle Data Network (EF4) DG ENV N°070201/2019/811467/ AA/ENV.B.1. Declarations Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Ardente F, Mathieux F (2014)	2021-04-21T14:13:33.324Z	2021-04-20T00:00:00.000	{ "year": 2021, "sha1": "856317cf77263fafa0419c2e8ac5c22d4253c60b", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/s11367-021-01893-2.pdf", "oa_status": "HYBRID", "pdf_src": "MergedPDFExtraction", "pdf_hash": "06943486b3a0473dd46975ed7f05a16721af71b3", "s2fieldsofstudy": [ "Political Science" ], "extfieldsofstudy": [] }
271926994	pes2o/s2orc	v3-fos-license	60. Assessing Long Term Alloplastic Breast Reconstruction Efficacy In A Longitudinal Cohort Of 9,232 Patients deep Minji Kim, BS, Sameer Massand, MD, Joshua Barnett, MD, Uchechukwu Amakiri, BS, Lillian Boe, PhD, Robert J. Allen, Jr, MD, Carrie S. Stern, MD, Babak Mehrara, MD, Jonas A. Nelson, MD, MPH Memorial Sloan Kettering, New York City, NY, USA PURPOSE: Implant-based breast reconstruction (IBR) is the most common approach for postmastectomy reconstruction.However, no study has examined which IBR patients are more likely to have higher patient reported outcomes (PROs) beyond 2-years postoperatively.This study aims to determine which variables are associated with PROs at 5-years postoperatively and to assess the proportion of patients undergoing revision procedures. METHODS: Patients who underwent direct-to-implant and two-staged IBR between 2010 and 2022 and completed the BREAST-Q were included.Primary outcomes were the four BREAST-Q scores at 5-years.The minimally clinically important difference was a score difference of 4. Secondary outcomes were revision procedures, including breast revision, fat grafting, and nipple reconstruction. CONCLUSION: Satisfaction with Breasts may decrease 5-years after IBR.Patients of Hispanic ethnicity, current/ former smoking status, and undergoing radiation should be counseled about decreased long-term PROs and that many patients undergo additional procedures. Minji Kim, BS, Barkat Ali, MD, Kevin Zhang, BA, Perri Vingan, BS, Lillian Boe, PhD, Catherine L. Ly, MD, Robert J. Allen, Jr, MD, Evan Matros, MD, Peter G. Cordeiro, MD, Babak Mehrara, MD, Jonas A. Nelson, MD, MPH Memorial Sloan Kettering, New York City, NY, USA PURPOSE: The incremental impact of increasing age on complications and patient reported outcomes (PROs) after postmastectomy breast reconstruction (PMBR) remains unknown.The purpose of this study is to understand the impact of age on complications and PROs using BREAST-Q longitudinally 5-years after PMBR. CONCLUSION: This 5-year longitudinal analysis of surgical outcomes and PROs suggests that complication rates increase with age and older age has a negative correlation with Physical Well-being of the Chest and Satisfaction with Breasts but a positive correlation with Psychosocial Wellbeing, regardless of the reconstructive type.Overall, age should be considered alongside other important factors, such as frailty, when determining whether a patient is a suitable candidate for PMBR. R'ay Fodor, BASc, Riley Marlar, DO, Ying Ku, BS, Mazen Al-Malak, MD, Jacob Lammers, DO, Lianne Mulvihill, BA, Diane Jo, MA, Ryan Khalaf, BS, Jose Reyes, BA, Fuad Abbas, BS, Antonio Rampazzo, MD, PhD, Bahar Bassiri Gharb, MD, PhD Cleveland Clinic, Cleveland, OH, USA PURPOSE: Ehlers Danlos Syndrome (EDS) is considered a relative contraindication to elective plastic surgery, but data regarding surgical risks are inconsistent.We hypothesized that the plastic surgical complication rate of EDS patients would not be significantly different compared to matched controls. METHODS: A retrospective matched cohort study of EDS patients who underwent plastic surgery at a tertiary healthcare system from 2003-2022 was performed.Controls were based on procedure, sex, age, comorbidities, and medications affecting wound healing.	2024-08-24T05:27:32.495Z	2024-05-01T00:00:00.000	{ "year": 2024, "sha1": "011ed4413657f914ec64a061a6c986eed4e9876e", "oa_license": "CCBYNCND", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "011ed4413657f914ec64a061a6c986eed4e9876e", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [] }
4786343	pes2o/s2orc	v3-fos-license	Tricuspid Stenosis: A Rare and Potential Complication of Ventricular Septal Occluder Device Asymmetrical septal occluder device (ASOD) has made percutaneous closure of ventricular septal defect an easy and effective management option. Although there are reports of aortic and tricuspid valvular regurgitation after deployment of ASOD, only few cases of tricuspid stenosis (TS) has been reported so far in the literature. We report a case of malaligned ASOD that occurred after successful device closure resulting in TS along with mild tricuspid and aortic regurgitation requiring surgical retrieval. Transesophageal echocardiography played crucial role in detecting the cause of tricuspid valve dysfunction besides providing continuous monitoring during the procedure. We intend to emphasize the need of echocardiographic evaluation of the tricuspid valvular apparatus and aortic valve during and after the device deployment even after the successful device closure to prevent this rare complication. Introduction Ventricular septal defects (VSDs), the 2 nd most common congenital heart disease (CHD) only after the bicuspid aortic valve, is often treated using percutaneous techniques. [1] With the development of newer asymmetrical septal occlude device (ASOD), many VSDs which were thought unsuitable for percutaneous closure are being treated using these devices with acceptable morbidity and very low mortality. [2][3][4][5][6] Sometimes, after deployment of ASOD, tricuspid valve regurgitation (TR) and/or aortic valve regurgitation (AR) develop due to the proximity of the aortic valve and tricuspid valve to the perimembranous VSD. The occurrence of tricuspid stenosis (TS) is very rare. [6] We report an unusual case of TS along with mild TR and AR in a patient after successful device closure of a perimembranous VSD that required surgical retrieval of ASOD. Transesophageal echocardiography (TEE) after device retrieval showed ruptured chordae tendenae of septal leaflets of TV causing TR. Case Report A 12-year-old male child presented to our institution with a complaint of respiratory This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. For reprints contact: reprints@medknow.Com distress while playing. His general physical examination was unremarkable. Auscultation of chest revealed a grade 4/6 pan-systolic murmur over precordium. Transthoracic echocardiogram (TTE) showed an 8 mm perimembranous VSD with left to right shunt, normal biventricular function, and no TR. The gradient across VSD was 40 mmHg with no evidence of pulmonary arterial hypertension and Qp/Qs ratio >1.5:1. He underwent percutaneous device closure for VSD under general anesthesia with endotracheal intubation after confirmation of VSD suitability for same on TEE and cath study. A 10 mm ASOD (Cera TM membrane VSD occluder, Life tech scientific, China) was placed across the VSD through femoral approach, and the device was deployed in place after confirming correct position on TEE and angiography. Postdeployment angiography confirmed the position of ASOD across the VSD with minimal flow across it and normally functioning tricuspid and aortic valve [ Figure 1]. After the procedure, anesthesia was reversed, and tracheal tube was extubated while avoiding coughing and bucking. On post-ASOD placement day 1, TTE revealed the device located in the tricuspid valve inlet causing turbulent flow across tricuspid valve and mild aortic Videos Available on: www.annals.in Figure 2b]. Clinically, the patient remained asymptomatic and hemodynamically stable. A decision for surgical removal of the ASOD and closure of the VSD was taken. In the operating room, after establishing intravenous access, ASA standard monitoring and invasive arterial monitoring, anesthesia was induced with titrated dose of propofol and fentanyl. Injection vecuronium was used to facilitate endotracheal intubation. The maintenance of anesthesia was achieved with inhalation of isoflurane, fentanyl 1 µg/kg/h and intermittent boluses of vecuronium. A TEE probe (iE 33, Philips Ultrasound, Bothell WA, USA) was inserted into the esophagus after securing the airway. Central venous cannulation was done through the left internal jugular vein with continuous monitoring under TEE to reduce the possibility of ASOD displacement by the guide wire or catheter. Intraoperative TEE confirmed the TTE finding of tricuspid inflow obstruction and showed the presence of mild TR [ Figure 3a and b and Video 2]. Transgastric inflow view confirmed the gradient across tricuspid valve (peak/mean 7/3 mmHg) [ Figure 3c]. Mid-esophageal 4 chamber view and modified 4 chamber view with retroflextion of probe showed that the device is malalinged such that the right ventricular rim of the device override the septal tricuspid leaflet and encroached on the anterior and posterior tricuspid valve leaflets resulting in improper opening and closing of the valve [ Figure 4a, b and Videos 3 and 4]. Cardiopulmonary bypass (CPB) was instituted using bicaval cannulation under continuous monitoring of TEE to prevent potential ASOD embolization. Intraoperatively, the surgeons found the ASOD stuck in the tricuspid valve. The device was partly entangled in the margin of anterior tricuspid leaflet The device was retrieved with utmost care to prevent further damage to the leaflets, chordae tendineae, and the conduction system. Visual inspection of the tricuspid valve after removal of device showed tear in the anterior leaflet and rupture chordae tendenae of septal leaflet. The anterior tricuspid leaflet was repaired with primary suturing. The VSD was closed with poly tetra flouro-ethylene patch. CPB was terminated using nor-adrenaline 0.05 µg/kg/min. TEE after termination of CPB revealed no residual VSD; the pulse wave Doppler across tricuspid valve showed no gradient (3/1 mmHg) and ruptured chordae tendineae of septal leaflet causing eccentric TR jet [ Figure 5 and Video 5]. After surgery, the patient was shifted to intensive care unit for elective mechanical ventilation which was weaned off after 4 h. He was discharged home after an uneventful stay of 5 days in the hospital. Discussion Isolated VSD comprises 20% of all CHD. [7] The development of congestive heart failure, pulmonary over-circulation causing recurrent respiratory tract infection, and pulmonary hypertension necessitates the closure of these VSD once detected with few exceptions like small muscular VSD which may close spontaneously as the child grows. [8] Surgical closure of VSD is associated with complications like residual defects, damage to the adjacent tricuspid valve, aortic valve, and conduction systems besides inflammatory and stress response of CPB on various organ system. Despite advancement in the surgical and CPB techniques reducing the above-mentioned complications, the ease of closure and the avoidance of CPB-related complication make the percutaneous technique a preferred approach over surgical closure. First attempt at closing a VSD using device was done using Rashkind umbrella device designed for closing patent ductus arteriosus. [9] With the development of ASOD, most of the isolated VSD are now successfully closed in the catheterization laboratory. The need of the adequate rim for the placement of these ASODs makes perimembranous and muscular VSD suitable for this approach. Since in our case the defect was perimembranous with adequate rims we planned for ASOD closure. Even though the success rate of complete closure using the ASOD is high (>90%), [10] it is not without risks. In the immediate postdeployment period, device malposition and embolization often occur while in the long-term conduction block, injury to the valvular structure causing TR and/or AR, left ventricular outflow tract obstruction and hemolysis may occur. [1,[11][12][13][14][15][16] The incidence of postdevice deployment conduction blocks vary from 1% to 5%. [5,6,17] TR and AR are not infrequent due to the associated damage to these valvular apparatus during device deployment. [3,6,18] Mertens et al. [18] reported severe TR 1 month after the successful device closure of the VSD. On retrospective analysis of the cine imaging, they noted the entrapment of the chordae into the device during deployment which did not cause much problem in the immediate postprocedure period whereas the repeated stress on the device screw assembly due to each cardiac contraction may have led to metal fatigue leading to its rupture and entrapment of the ruptured segment in the chordal structures. The TS is a rare and serious complication that requires attention in the postprocedure period. Till now only two cases of device-induced TS has been reported in the literature, Arora et al. [6] reported a case of TS in the study of a large group of VSD device closure patients that was relieved by balloon dilatation. Christiani et al., [10] found TS immediately following postdevice deployment due to entrapment of the right ventricular disc in the chordae of the anterior tricuspid leaflet which was relieved by chance when they tried to retrieve the device. They proposed that right coronary Judkin catheter that was passed through the defect from the arterial side would have passed through the chordae of the anterior leaflet of tricuspid valve given its proximity to the VSD, thus causing the entrapment of the right ventricular disc into the chordal structure during the device deployment. They suggested that device entrapment in the adjacent structures may be suspected whenever the ventricular disc does not get its prescribed shape after deployment of the device and to try manipulation of the adjacent structures such as tricuspid valvular apparatus to relieve the device entrapment before approaching surgical retrieval; however, danger of device displacement and embolisation should be kept in mind before manipulation. The device malalignment may be fatal at times as in the Christiani et al. [10] report, TS caused pressure elevation in the right atrium leading to shunting of blood through patent foramen ovale (PFO) into the left atrium causing desaturation and relief of right atrial pressure, however, in the absence of PFO same incident may have been fatal before detection of the cause. In our case, immediately after deployment cine study showed the normal shape and placement of the device with the closure of the VSD, however, postdevice placement day 1 TTE showed device in the tricuspid valve inlet causing TS. TEE confirmed the TTE finding and showed that TS was due to malalignment of the device. Probably in our case, part of tricuspid valve chordae got entrapped in the device during deployment that was not significant enough to cause tricuspid valve functional abnormality at that time. With the repeated cardiac contraction, the device may have got entangled into the tricuspid valve apparatus causing malalignment resulting in TS and damage to the valve leaflets due to friction force caused by repeated movement of valve on the ASOD. Few strategies suggested to prevent entrapment of chordal structure in the device includes straightening of the wire after snaring to look for any kinking in the wire, monitoring for TR or TS during the passage of the device delivery sheath across the VSD and using curved catheter to cross the VSD from left ventricular side in place of Judkin catheter thus reducing the chance of wire passing through the chordae. [18] In our case, although Judkin catheter was used there was no signs of abnormal flow pattern across tricuspid valve after device deployment, suggesting that entrapment of the tricuspid valve apparatus was not severe enough to produce a change in the inflow pattern in the immediate postprocedure period. Once the complications are diagnosed, the best treatment strategy should be surgical removal of the device and closure of the defect, as an attempt to remove through percutaneous technique may lead to adverse effects such as damage to tricuspid valve apparatus, adjacent conduction system, and device embolization. This case report adds to the importance of echocardiographic evaluation of the tricuspid valve apparatus both during the device deployment, in the immediate postdeployment period and during long-term follow-up. Conclusion Tricuspid valve stenosis although rare may present few days after device deployment warranting monitoring before discharge from the hospital. A special attention should be given to avoid entanglement of device in the tricuspid valve chordal structures to avoid this complication. TEE may play crucial role to find the cause of valve dysfunction after device deployment. Declaration of patient consent The authors certify that they have obtained all appropriate patient consent forms. In the form the patient(s) has/have given his/her/their consent for his/her/their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.	2018-04-26T23:46:28.726Z	2018-04-01T00:00:00.000	{ "year": 2018, "sha1": "874ebbe39e75a094e475a006d876f8ac9e6d280f", "oa_license": "CCBYNCSA", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "874ebbe39e75a094e475a006d876f8ac9e6d280f", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
119710540	pes2o/s2orc	v3-fos-license	Comparing localizations across adjunctions We show that several apparently unrelated formulas involving left or right Bousfield localizations in homotopy theory are induced by comparison maps associated with pairs of adjoint functors. Such comparison maps are used in the article to discuss the existence of functorial liftings of homotopical localizations and cellularizations to categories of algebras over monads acting on model categories, with emphasis on the cases of module spectra and algebras over simplicial operads. Some of our results hold for algebras up to homotopy as well; for example, if $T$ is the reduced monad associated with a simplicial operad and $f$ is any map of pointed simplicial sets, then $f$-localization coincides with $Tf$-localization on spaces underlying homotopy $T$-algebras, and similarly for cellularizations. Introduction Preservation of structures such as loop spaces, infinity loop spaces, or module spectra under homotopical localizations or cellularizations has been studied using Segal's theory of loop spaces [15,33], operads [26,46,76,77,78], algebraic theories [4,13], or other methods [16,18,28,47]. Monads and their algebras lie behind many of these approaches. However, although the existence of liftings of localizations or colocalizations to categories of algebras over monads has been proved in various special cases, functoriality of such liftings has only been addressed recently in [9,48,79] as well as in the present article. This article emerged from the observation that some formulas involving localizations or cellularizations in homotopy theory share common patterns not previously revealed. The formulas that we consider -some of which are well known while others are new-contain pairs of adjoint functors in some way or another. As a first example, it was shown by Farjoun in [33,Theorem 3.A.1] that for every pointed connected space X and every basepoint-preserving map f there is a weak equivalence where L f denotes localization with respect to f in the homotopy category of pointed spaces, Ω is the loop functor, and Σ denotes suspension. Similarly, Bousfield proved in [16, Theorem 2.10] that for every spectrum X and every pointed connected space A one has where P A is A-nullification (that is, localization with respect to the map A → * ) and Σ ∞ is the canonical functor from the homotopy category of pointed spaces to the homotopy category of spectra, while Ω ∞ is its right adjoint. In Section 7 we prove that, as one would expect, the formula (1. for every basepoint-preserving map f and every commutative topological monoid X, where SP ∞ is the infinite symmetric product [32]. The stable analogue of this fact is the statement that, if HZ denotes the spectrum that represents homology with integral coefficients, then (1.4) L f X ≃ L HZ∧f X for every f and every spectrum X splitting as a product of Eilenberg-Mac Lane spectra. Another seemingly unrelated fact was observed in [24,Corollary 4.3], namely that there is a natural homomorphism of groupoids (1.5) πL f X −→ L πf (πX) for all spaces X and every map f , where π is the fundamental groupoid. This homomorphism is almost never an isomorphism, yet it becomes an isomorphism after applying L πf to it. Further details about these examples and many others are given in the article. Our encompassing approach is based on a study of comparison maps of type (1.6) α : F L f −→ L F f F and β : L f G −→ GL F f for each Quillen pair of adjoint functors F and G between model categories. In fact α and β form a pair of mates as in [74]. Such comparison maps arise very frequently in practice but are equivalences only in favorable cases. It is remarkable that α becomes an equivalence after applying L F f to it, as in (1.5), while β does not share this feature. The formulas (1.1), (1.2), (1.3) and (1.4) are "of β type" while (1.5) is "of α type". Among the new formulas that we obtain by means of comparison maps we emphasize the following ones. For a cofibrant ring spectrum E and a map f of spectra, if f -localization commutes with suspension or E is connective, then there is a natural equivalence The content of the article may be summarized as follows. 1.1. Comparison maps. We start with a purely categorical study of natural transformations such as α and β in (1.6), relating localizations or colocalizations defined in two distinct categories linked by pairs of adjoint functors. This is the core technique of the article, dating back to a preliminary draft written in 2006, which was gradually adapted for its use in a homotopy-theoretical context. Our task was made feasible by results about transferred model structures on categories of algebras over monads (or algebras over operads) from articles that appeared in the meantime. 1.2. Preservation of (co)algebras under (co)localizations. Our main result in Section 4 provides necessary and sufficient conditions under which a localization on a category preserves algebras over a monad defined on the same category. This result uses comparison maps (1.6) associated with the Eilenberg-Moore adjunction of the given monad. The central finding, inspired by results in [28], is that, if a monad T and a localization L are defined on the same category, then T preserves L-equivalences if and only if L preserves T -algebras. This fact is made precise in Theorem 4.2 and is reminiscent to results in [11], where distributivity between monads was first discussed. Theorem 4.2 can be dualized in two different ways. One obvious way consists of passing to opposite categories -this yields conditions under which colocalizations preserve coalgebras over comonads. The other way is achieved by means of orthogonality between objects and morphisms, leading to conditions under which colocalizations preserve algebras over monads or localizations preserve coalgebras over comonads. 1.3. Lifting (co)localizations to model categories of algebras. We subsequently address a homotopy-theoretical version of the results in Section 4, hence providing necessary and sufficient conditions under which a localization or a colocalization on a model category M lifts to a category of algebras M T over a monad T acting on M, assuming that the category M T admits a model structure transferred from M; that is, one for which the forgetful functor U : M T → M creates weak equivalences and fibrations. Existence of such transferred model structures (or semi model structures) has been discussed in several articles, such as [8,12,65,75]. Our main results in this part of the article (Theorem 7.4 and its dual counterpart, Theorem 7.8) were obtained in collaboration with Javier Gutiérrez and David White, who arrived at similar conclusions in [9,48,77]. It follows from results in Section 7 below or from [9] that a homotopical f -localization lifts to a category of T -algebras admitting a transferred model structure if and only if the forgetful functor U sends F f -equivalences to f -equivalences, where F is Quillen left adjoint to U , assuming that L F f exists -in fact, in this case L F f is a lift of L f to T -algebras. Similarly, A-cellularization lifts to T -algebras if and only if the forgetful functor U sends F A-cellular objects to A-cellular objects, and in this case C F A is a lift of C A to T -algebras, assuming the existence of C F A . We apply Theorem 7.4 and Theorem 7.8 to modules over a cofibrant ring spectrum E and to algebras over simplicial operads. In both cases, transferred model structures are known to exist. However, there is a crucial distinction: in the case of E-module spectra the associated monad T X = E ∧ X preserves all colimits, while in the case of algebras over a simplicial operad P the corresponding monad (whose algebras are the P -algebras) only preserves sifted colimits. Therefore our treatment of the case of module spectra is easier. 1.4. Localizations and cellularizations of module spectra. We infer that, if E is a cofibrant ring spectrum (in the category of symmetric spectra over simplicial sets), then f -localizations and A-cellularizations lift to the category of left E-modules assuming that they commute with suspension or that E is connective. This was first shown in [26,46] by viewing E-module spectra as algebras over a suitable two-colour operad. Furthermore, if L f (or C A ) commutes with suspension or E is connective, then for every left E-module M there are natural equivalences 10) that explain and refine (1.7) and (1.8). Statements and proofs of claims about cellularizations turn out to be formally analogous to those about localizations, since passage from L f to C A involves exchanging objects with morphisms or conversely. However, as our results are formulated in terms of right-induced model structures on categories of algebras over monads, cellularizations tend to behave better than localizations. For example, right Bousfield localizations with respect to objects commute with forming Eilenberg-Moore categories of monads (Theorem 7.13 below) while left Bousfield localizations with respect to morphisms need not do so (Example 8.4). 1.5. Algebras over simplicial operads. In Section 9 we recover and extend results of Bousfield and Farjoun about interaction of localizations or cellularizations with loop spaces and infinite loop spaces, such as (1.1) and (1.2). For this, we prove that Theorem 7.4 and Theorem 7.8 apply to model categories of algebras over simplicial operads acting on pointed simplicial sets. Let us emphasize that our main result in this section (Theorem 9.6) not only states that localizations and cellularizations preserve algebras over simplicial operads, but they do it in a functorial way, that is, yielding localization functors and cellularization functors on algebras. In this respect, our approach shares insight with [48]. 1.6. Algebras up to homotopy. Examples hinted that some of the above conclusions did not really require the existence of transferred model structures on categories of algebras over monads, but held equally well for algebras up to homotopy, i.e., algebras over the derived monad in the homotopy category. For example, the equivalences L f M ≃ L E∧f M and C A M ≃ C E∧A M still hold when M is a homotopy E-module. More generally, we found that if X underlies a homotopy T -algebra, provided that T preserves f -equivalences and T f -equivalences, or, correspondingly, A-cellular spaces and T A-cellular spaces. These assumptions on T are automatically satisfied if T is the reduced monad associated with a simplicial operad acting on pointed simplicial sets. Our proof of this fact relies on arguments used by Farjoun in related parts of [33]. As consequences we infer that assuming in each case that X is a homotopy algebra over the corresponding monad, namely ΩΣ in the first case (whose algebras are the homotopy associative H-spaces) and Ω ∞ Σ ∞ in the second case (whose algebras are called H ∞ -spaces). The second one can be generalized as for algebras over the monad T X = Ω ∞ (E ∧Σ ∞ X), where E is any connective ring spectrum. This is an unstable analogue of the equivalence L f X ≃ L E∧f X for homotopy E-modules, and if E = HZ then one recovers the formula (1. All these formulas remain valid if L f is replaced with cellularization C A with respect to some pointed connected space A. Most of our results are true, more generally, for localizations with respect to collections of maps (possibly proper classes) and for colocalizations with respect to collections of objects, whenever these exist. Our results in Section 7 are conceptually related with the content of [20, § 4], where it is shown that the derived localization of a differential graded algebra A at a set of homogeneous homology classes coincides with the derived localization of A as a left A-module. Equivalences such as (1.9) and (1.10) probably hold, with suitable assumptions on E, for other stable model categories, for example motivic symmetric spectra over any base scheme B, assuming the existence of Bousfield localizations -this is the case when B is Noetherian and of finite Krull dimension [37]. In this context, (1.10) may be relevant in the study of the derived category DM(k) of motives over a field k of characteristic zero, which is equivalent, according to [69], to the homotopy category of modules over the spectrum M Z that represents motivic cohomology for the given field k. Note that (1.10) with A = S (the motivic sphere spectrum) implies that E is S-cellular, which is a strong restriction [36]. Hence, S-cellularity of E is necessary for (1.10) to hold for all A, perhaps also sufficient. Adjunctions, monads and comonads This section contains standard terminology and basic facts that will be used in the article. More information can be found in [57,Chapters IV and VI]. If C and D are categories, we denote by F : C ⇄ D : G a pair of adjoint functors, with F left adjoint and G right adjoint, meaning that there are natural bijections of morphism sets for X in C and Y in D. We denote by ϕ t : X → GY the adjunct of a morphism ϕ : F X → Y under (2.1) and, similarly, ψ t : F X → Y denotes the adjunct of ψ : X → GY . Adjuncts of identities yield natural transformations η : Id → GF (called unit ) and ε : F G → Id (called counit ), which, in turn, determine the adjunction by ψ t = ε Y • F ψ and ϕ t = Gϕ • η X . Passage to opposite categories transforms an adjunction F : C ⇄ D : G into another adjunction G : D op ⇄ C op : F , where G is now the left adjoint and the former unit becomes the counit, and conversely, since C(X, GF X) = C op (GF X, X) for all X. If F : C ⇄ D : G is a pair of adjoint functors, then F is a retract of F GF and G is a retract of GF G. This follows from the triangle identities for all objects X and Y . Moreover, if we consider the full subcategories then F and G restrict to an equivalence of categories A monad on a category C is a triple (T, η, µ) where T : C → C is a functor and η : Id → T (the unit ) and µ : T T → T (the multiplication) are natural transformations such that A monad (T, η, µ) is called idempotent if µ is an isomorphism, which we then omit from the notation. If F : C ⇄ D : G is an adjunction with unit η and counit ε, then (GF, η, GεF ) is a monad. In fact, all monads are of this form, in a non-unique way. If (T, η, µ) is a monad on a category C, then a T -algebra is a pair (X, a) with a : T X → X such that a • T a = a • µ X and a • η X = id X . A morphism of T -algebras (X, a) → (Y, b) is a morphism ϕ : X → Y in C such that ϕ • a = b • T ϕ. Thus, the T -algebras form a category, denoted by C T and called the Eilenberg-Moore category of T , which fits into an adjunction where F X = (T X, µ X ) and U (X, a) = X. This adjunction is terminal among all adjunctions whose associated monad is T . A comonad on a category C is a monad on the opposite category C op . We denote a comonad by (T, ε, δ) where ε : T → Id is the counit and δ : T → T T is the comultiplication. A coalgebra over (T, ε, δ) is a pair (X, a) with a : X → T X such that T a • a = δ X • a and ε X • a = id X . If F : D ⇄ C : G is an adjunction with unit η and counit ε, then (F G, ε, F ηG) is a comonad. The category C T of T -coalgebras provides an initial adjunction U : C T ⇄ C : G among those yielding T , where GX = (T X, δ X ) for all X; cf. [51, § 2]. A full subcategory S of a category C is reflective if the inclusion J is part of an adjunction K : C ⇄ S : J. In this case, the counit KJ → Id is an isomorphism and the functor L = JK is called a reflector or a localization on C. If we denote the unit by l : Id → L, then (L, l) is an idempotent monad. An object of C is called L-local if it is isomorphic to an object in the subcategory S; hence, X is L-local if and only if l X : X → LX is an isomorphism. A morphism g : U → V is an L-equivalence if Lg is an isomorphism, or, equivalently, if for all L-local objects X composition with g induces a bijection Conversely, the L-local objects are precisely those X for which (2.5) holds for all L-equivalences g : U → V ; see [2] for further details. Dually, a full subcategory S of a category D is coreflective if the inclusion J has a right adjoint J : S ⇄ D : K. Then the functor C = JK is called a coreflector or a colocalization on D, and in this case the unit Id → KJ is an isomorphism. If we denote the counit by c : C → Id, then (C, c) is an idempotent comonad. An object of D is called C-colocal if it is isomorphic to an object in S; thus, X is C-colocal if and only if c X : CX → X is an isomorphism. A morphism g : U → V is a C-equivalence if Cg is an isomorphism, or, equivalently, if for all C-colocal objects X composition with g induces a bijection The C-colocal objects are precisely those X for which (2.6) holds for all C-equivalences g : U → V . In fact, colocalizations are localizations on the opposite category. Comparison morphisms Suppose given a localization L 1 on a category C 1 and a localization L 2 on a category C 2 . We say that a functor F : C 1 → C 2 preserves local objects if F X is L 2 -local for every L 1 -local object X, and we say that F preserves equivalences if F f is an L 2 -equivalence whenever f is an L 1 -equivalence. We also say that a functor F : C 1 → C 2 reflects local objects if, for an object X of C 1 , the assertion that F X is L 2 -local implies that X is L 1 -local. Similarly, F reflects equivalences if f is an L 1 -equivalence whenever F f is an L 2 -equivalence. The same terminology will be used for colocalizations. Proof. This follows from the definitions, in view of the commutative diagram for a morphism A → B in C 1 and an object X in C 2 , where the vertical bijections are given by the adjunction. Part (b) is deduced from part (a) by passing to the opposite categories. Theorem 3.2. Let F : C 1 → C 2 be a functor. Let L 1 be a localization on C 1 with unit l 1 and L 2 a localization on C 2 with unit l 2 . Then the following hold: (i) F preserves equivalences if and only if there is a natural transformation Proof. For every X in C 2 , the morphism (l 1 ) X : X → L 1 X is an L 1 -equivalence. Therefore, if F preserves equivalences, then F (l 1 ) X is an L 2 -equivalence. Hence it induces a natural bijection and α X is uniquely defined by the equality α X • F (l 1 ) X = (l 2 ) F X . Since (l 2 ) F X and F (l 1 ) X are both L 2 -equivalences, α X is also an L 2 -equivalence. In order to prove that α is a natural transformation, we need to check that This follows from the equality using the fact that F (l 1 ) X is an L 2 -equivalence and L 2 F Y is L 2 -local. Next we assume that there is a natural transformation α such that α•F l 1 = l 2 F and infer that F preserves equivalences. For this purpose, let f : In order to prove that F f is an L 2 -equivalence it suffices to check that s is an inverse of L 2 F f . This is deduced from the equalities If F preserves equivalences, then the equality α X • F (l 1 ) X = (l 2 ) F X implies that α X is an L 2 -equivalence, and therefore α X is an isomorphism if and only if F L 1 X is L 2 -local. Hence α is an isomorphism if and only if F sends L 1 -local objects to L 2 -local objects. This completes the proof of part (i). The proof of part (ii) is similar. For a localization L 1 on C 1 and a localization L 2 on C 2 , the functors F L 1 and L 2 F are naturally isomorphic if and only if F preserves local objects and equivalences. In this case, α : F L 1 → L 2 F and β : L 2 F → F L 1 are mutually inverse isomorphisms. Proof. The "if" part follows from Theorem 3.2. For the converse, note that if F L 1 ∼ = L 2 F then F preserves local objects, and the naturality of the isomorphism adds the fact that F preserves equivalences, since, for a morphism f , we have that L 2 F f is an isomorphism if and only if F L 1 f is an isomorphism. Furthermore, if F preserves local objects and equivalences, then the equality α • β • l 2 F = l 2 F implies that α • β = id. Corollary 3.4. Suppose that a functor U : C 1 → C 2 reflects isomorphisms. For a localization L 1 with unit l 1 on C 1 and a localization L 2 with unit l 2 on C 2 , if U L 1 and L 2 U are naturally isomorphic then the following hold: (i) U preserves and reflects local objects and equivalences. (ii) There is a natural isomorphism α : U L 1 → L 2 U such that α • U l 1 = l 2 U , and a natural isomorphism β : L 2 U → U L 1 such that β • l 2 U = U l 1 . Moreover, the isomorphisms in (ii) are unique and inverse to each other. Proof. Corollary 3.3 tells us that U preserves local objects and equivalences, and that α : U L 1 → L 2 U and β : L 2 U → U L 1 , as given by Theorem 3.2, are inverse isomorphisms. To prove that U reflects local objects, suppose that U X is L 2 -local. Then (l 2 ) UX is an isomorphism. Since α • U l 1 = l 2 U and α X is an isomorphism, we infer that U (l 1 ) X is an isomorphism and therefore so is (l 1 ) X since U reflects isomorphisms. Hence, X is L 1 -local, as needed. The fact that U reflects equivalences follows from the equality for every g : X → Y , together with the fact that U reflects isomorphisms. Example 3.5. For a cocomplete category C and a small category I, choose F to be the colimit functor colim I : C I → C, where C I denotes the category of functors I → C. Let L be a localization on C and extend it objectwise over C I ; that is, for each X : I → C, define (LX) i = L(X i ) for all i ∈ I. Then the diagonal functor C → C I preserves local objects and equivalences. Since it preserves local objects, the colimit functor preserves equivalences by Lemma 3.1. Therefore Theorem 3.2 yields a natural morphism This is an instance of the well-known fact that left adjoints preserve colimits. Since the diagonal functor C → C I also preserves equivalences, if C is complete then the limit functor preserves local objects. This yields a natural morphism which is rarely an equivalence of any type. However, for every colocalization C there is an for each X : I → C, as a special case of the next result, which follows from Theorem 3.2 by passing to the opposite categories. The duals of Corollary 3.3 and Corollary 3.4 hold as well. As a special case of Corollary 3.3 and its dual, we obtain conditions for commutativity of a localization and a colocalization. Corollary 3.7. If L is a localization and C is a colocalization on the same category, then the following statements are equivalent: (a) There is a natural isomorphism LC ∼ = CL. (b) L preserves C-colocal objects and C-equivalences. (c) C preserves L-local objects and L-equivalences. Induced (co)localizations on (co)algebras If often happens that a localization or a colocalization preserves a certain subcategory. For example, every localization on the category of groups preserves abelian groups [22]. As we next point out, in such cases the restriction is also a localization or a colocalization. The claim that an endofunctor F on a category C preserves a subcategory S means that F X is in S for every object X of S and F f is in S for every morphism f of S. If S is full, then the second condition is implied by the first one. (a) If a localization L on C preserves S then L restricts to a localization on S, and the inclusion S ֒→ C preserves and reflects local objects and equivalences. (b) If a colocalization C on C preserves S then C restricts to a colocalization on S, and the inclusion S ֒→ C preserves and reflects colocal objects and equivalences. Proof. For (a), consider the full subcategory L of S consisting of all L-local objects of C that are in S. Then, for each object X in S, the morphism l X : X → LX is in S by assumption, and, for each Y in L, it induces a bijection so L restricts indeed to a reflection of S onto L such that the inclusion preserves and reflects local objects. If f : X → Y is a morphism in S, then Lf is an isomorphism in S if and only if it is an isomorphism in C, since the inclusion reflects isomorphisms. Hence, the inclusion also preserves and reflects equivalences. Passage to the opposite category yields (b). In a similar spirit, the following result enhances [28, Theorem 1.2] by adding the remark that, if a localization sends algebras over a monad to algebras over the same monad, then it yields in fact a localization on the category of such algebras. The noteworthy principle contained in this theorem is that, for a monad T and a localization L on the same category, T preserves L-equivalences if and only if L preserves T -algebras. Proof. We first show that (a) ⇒ (b). To obtainã : T LX → LX withã • T l X = l X • a, we use the fact that LX is L-local and T l X is an L-equivalence by assumption, and note thatã is then unique. The relationã • Tã =ã • µ LX follows, as in the proof of [28, Theorem 1.2], from the equalitỹ since T T l X is an L-equivalence. Similarly, we infer thatã • η LX = id LX from the equalitỹ a • η LX • l X = l X . Given a morphism of T -algebras g : (X, a) → (Y, b), we find that which implies that Lg •ã =b • T Lg, as claimed. Next we prove that (b) ⇒ (c). For each T -algebra (X, a), let us define L T (X, a) = (LX,ã), whereã is given by assumption. Thus, for every X and all a : T X → X. Moreover, we set l T = l, so we have indeed lU = U l T . For a morphism of T -algebras g : (X, a) → (Y, b), we define L T g = Lg, which is a morphism of T -algebras by assumption. Hence U L T g = LU g, and L T is a functor because so is L. To check that L T is a localization, suppose given any morphism g : (X, a) → (Y, b) of T -algebras, and suppose that Y is L-local. Then there is a unique g ′ : LX → Y in C such that g ′ • l X = g, and the equality l Y • g ′ • l X = Lg • l X implies that l Y • g ′ = Lg as well. We need to prove that g ′ is also a morphism of T -algebras, that is, that g ′ •ã is equal to b • T g ′ . For this, we use the fact that Lg is a morphism of T -algebras by assumption to infer that and then we use the fact that l Y is an isomorphism since Y is L-local. The implication (c) ⇒ (d) is trivial. Finally, we prove that (d) ⇒ (a). Suppose that a natural isomorphism LU ∼ = U L T is given. Then Corollary 3.4 tells us that U preserves and reflects local objects and equivalences. Now write T = U F where F X = (T X, µ X ), and note that F preserves equivalences by Lemma 3.1. Since U also preserves equivalences, so does T , hence yielding (a). When condition (c) of Theorem 4.2 is satisfied, we say that L T is a lifting of L to the category C T of T -algebras. This notion was already considered by Beck in [11], where it was shown that the existence of a lifting of L to C T is in fact equivalent to the existence of a distributive law of T over L, that is, a natural transformation λ : T L → LT subject to the conditions λ • ηL = Lη, λ • T l = lT , and λ • µL = Lµ • λT • T λ (the fourth condition in [11] is automatic in our case since L is idempotent). Under these conditions, LT becomes a monad by means of λ. Indeed, it follows from part (i) of Theorem 3.2 that, if L T is a lifting of L to C T , then there is a unique natural transformation The condition that T preserves L-equivalences holds for all localizations L whenever C(T −, −) = Φ • C(−, −) for some functor Φ, as in the next example. Example 4.4. Let Ho(Sp) be the homotopy category of spectra and let E be a homotopy ring spectrum, that is, a monoid in Ho(Sp). Then T X = E ∧ X defines a monad on Ho(Sp), whose algebras are left homotopy E-module spectra. If A is any spectrum, then an A * -equivalence is a map of spectra f : . It was proved in [14] that there is an A * -localization functor (−) A on Ho(Sp) whose class of equivalences is precisely the class of A * -equivalences. Since T preserves A * -equivalences, it follows from Theorem 4.2 that (−) A lifts to the category of E-module spectra, as already pointed out in [18,25]. The resulting functor annihilates precisely the is zero and, since it is also the identity map, we conclude that A ∧ M = 0. This argument yields that If the monad (T, η, µ) is idempotent, as in the next example, then the conditions of Theorem 4.2 are, in their turn, equivalent to the condition that L preserves the full subcategory of T -local objects, and in this case the lifting L T is just the restriction of L to this subcategory. Example 4.5. If L is any localization on the category of groups, then, as explained in [22,Theorem 2.2], L preserves abelian groups and hence restricts to the full subcategory of these. Therefore there is a natural group homomorphism and a natural isomorphism L((LG) ab ) ∼ = L(G ab ) for all groups G and every localization L. We note, however, that (4.3) is far from being an isomorphism in general. For instance, let P be a set of primes and let P ′ denote its complement. A group G is uniquely P ′ -divisible if the map x → x q is bijective in G for every q ∈ P ′ . If l G : G → G P denotes a universal homomorphism from G into a uniquely P ′ -divisible group, then (−) P is a localization on the category of groups whose restriction to abelian groups is tensoring with Z P . As shown in [10], if F is a free group of rank n ≥ 2 then ( As next shown, Theorem 4.2 also holds for coalgebras over a comonad. The proof is totally analogous, but the roles of local objects and equivalences are exchanged. Thus, the motto is now that a comonad T preserves L-local objects if and only if L preserves T -coalgebras. Theorem 4.6. Let (T, ε, δ) be a comonad on a category C and let L be a localization on C with unit l. Let U be the forgetful functor from the category C T of T -coalgebras to C. Then the following statements are equivalent: If these statements hold, then the T -coalgebra structuresã : LX → T LX given in part (b) are unique, and the localization L T in (c) is also unique. Moreover, U preserves and reflects local objects and equivalences. Proof. Suppose that T preserves L-local objects. In order to obtainã : LX → T LX with a•l X = T l X •a, use that T LX is L-local by assumption and l X is an L-equivalence, and note thatã is then unique. The relations Tã •ã = δ LX •ã and ε LX •ã = id LX are consequences of the equalities Tã •ã • l X = δ LX •ã • l X and ε LX •ã • l X = l X . The localization L T is defined by L T (X, a) = (LX,ã) and the rest of the proof follows the same steps as the proof of Theorem 4.2. The implication (d) ⇒ (a) follows from Lemma 3.1, as the right adjoint F X = (T X, ε X ) preserves local objects if and only if U preserves equivalences. By passing to opposite categories, Theorem 4.6 yields a result about preservation of algebras over a monad under the effect of a colocalization, which we next state for later reference, and Theorem 4.2 dualizes into a result relating colocalizations with coalgebras over a comonad, which we omit. Distributivity between comonads was previously studied in [6]. Distributive laws of monads over comonads have been considered in [50, § 2.3]. Theorem 4.7. Let (T, η, µ) be a monad on a category C and let C be a colocalization on C with counit c. Let U be the forgetful functor from the category C T of T -algebras to C. Then the following statements are equivalent: If these statements hold, then the T -algebra structuresã : T CX → CX given in part (b) are unique, and the colocalization C T in (c) is also unique. Moreover, U preserves and reflects colocal objects and equivalences. Proof. The given monad T defines a comonad on C op whose coalgebras are the T -algebras, and C defines a localization on C op whose local objects are the C-colocal ones. Thus all the statements follow from those in Theorem 4.6. If C is any colocalization on R-Mod, then T preserves C-colocal objects, since (4.1) yields for every C-colocal R-module N and every C-equivalence f between R-modules. Therefore, C lifts to the category S-Mod of left S-modules, as first shown in [43, Proposition 2.1]. Example 4.9. If C is any colocalization on the category of groups, then, according to [34], C preserves nilpotent groups of any nilpotency class k and hence restricts to the full subcategory of these. This yields a natural group homomorphism for every group G and all k, where Γ k denotes the kth term of the lower central series. For k = 1, the homomorphism (4.5) takes the form (CG) ab → C(G ab ), which is not an isomorphism in general, not even a C-equivalence. For example, the 3-torsion subgroup of the symmetric group Σ 3 is cyclic of order 3 while the abelianization of Σ 3 has no 3-torsion. Inverting morphisms and building from objects A collection of morphisms F and a collection of objects X in a category C are called orthogonal if for every morphism f : A → B in F and every object X in X the induced function is a bijection of sets. The objects orthogonal to F are called F -local and the morphisms orthogonal to the collection of all F -local objects are called F -equivalences. An F -localization of an object X is an F -equivalence l X : X → L F X into an F -local object. If an F -localization exists for all objects, then L F is indeed a localization on C. Note that, if a localization functor L is given, then L = L F where F is the collection of all L-equivalences. As shown in [1, Theorem 1.39], if a category C is locally presentable, then L F exists whenever F is a set, while if F is a proper class then the existence of L F can be proved if one assumes the existence of sufficiently large cardinals [1, Chapter 6]. Dually, a collection of morphisms F and a collection of objects A in a category C are co-orthogonal if for every morphism f : X → Y in F and every object A in A the induced function is a bijection of sets. The morphisms co-orthogonal to A are called A-equivalences and the objects co-orthogonal to the collection of all A-equivalences are called A-colocal. An A-colocalization of an object X is an A-equivalence c X : C A X → X from an A-colocal object into X. If an A-colocalization exists for every object, then C A is a colocalization on C. If a colocalization C is given, then C = C A where A is the collection of all C-colocal objects. If C is a locally presentable category, then the existence of C A is ensured for every set A by the dual of the Special Adjoint Functor Theorem [1, § 0.7], since the full subcategory of A-colocal objects is closed under colimits and has a generating set (namely A), and C is cowellpowered according to [1, Theorem 1.58]. If A is a proper class, then the existence of C A also follows from convenient large-cardinal axioms [1, Theorem 6.28]. Since the main source of motivation of the present article is the study of localizations of the form L f for a single morphism f and colocalizations of the form C A for a single object A, we will restrict statements of results to those cases, also for the sake of simplicity. However, most of our conclusions hold equally well for collections of morphisms instead of a single morphism, and for collections of objects instead of a single object, provided that the corresponding localizations or colocalizations exist. G be a pair of adjoint functors and let f : is an isomorphism if and only if F preserves local objects, and β is an isomorphism if and only if G preserves equivalences. Proof. For every object Y in C 2 , consider the commutative diagram where the vertical bijections are given by the adjunction and the horizontal arrows are induced by F f and f . It follows that GY is f -local if and only if Y is F f -local, as claimed. Part (ii) is a consequence of (i), and all the claims in (iii) follow from Theorem 3.2. The natural transformations α and β in part (iii) of Proposition 5.1 are mates; that is, each of them determines the other one as follows (see [74, p. 5]): where η is the unit of the adjunction and ε is the counit. The analogue of Proposition 5.1 for colocalizations reads as follows. G be a pair of adjoint functors and A an object in C 1 . There is also a unique natural transformation Proof. Parts (i) and (ii) are proved using the commutative diagram where the vertical bijections are given by the adjunction and the horizontal arrows are induced by g and Gg. Part (iii) comes from Theorem 3.6. Proposition 5.3. Let F : C ⇄ C T : U be the Eilenberg-Moore factorization of a monad T on a category C. Let f be a morphism in C such that L f exists. Then the following statements are equivalent: Proof. Suppose first that T preserves f -equivalences. Theorem 4.2 tells us that there is a localization L T on C T equipped with a natural isomorphism L f U ∼ = U L T , for which U preserves and reflects local objects and equivalences. Hence the L T -local objects are those Now assume that T preserves f -equivalences and T f -equivalences, and that L T f exists. Then the implication (i) ⇒ (ii) yields natural isomorphisms Since T preserves f -equivalences, T f is an f -equivalence. Hence all f -local objects are T f -local. If a T -algebra (X, a) is F f -local, then X is f -local by part (i) of Proposition 5.1, and hence X is T f -local. But this implies that (X, a) is F T f -local, again by part (i) of Proposition 5.1. Conversely, since T = U F , the fact that F is a retract of F U F by (2.2) implies that F f is an F T f -equivalence, and consequently every F T f -local T -algebra is F f -local. This allows us to conclude that the classes of F f -local T -algebras and F T f -local T -algebras coincide, so If T is an idempotent monad on a category C and we denote by J : S → C the inclusion of the full subcategory of T -local objects and by K : C → S its left adjoint, then, as a special case of Proposition 5.3, we infer, for a morphism f in C, the following facts: In fact, the proof is easier, since the counit ε of the adjunction is in this case an isomorphism and hence K ∼ = KJK. Example 5.4. As an example, let T be abelianization on the category of groups. Since all localizations on groups preserve abelian groups, part (ii) tells us that for every group homomorphism f and all abelian groups A. This fact was used in [28]. As in Section 4, there is an analogous version of Proposition 5.3 for colocalizations: Proposition 5.5. Let F : C ⇄ C T : U be the Eilenberg-Moore factorization of a monad T on a category C. Let A be an object in C such that C A exists. Then the following statements are equivalent: (i) T preserves A-colocal objects. (ii) C F A exists and there is a natural isomorphism Moreover, if T preserves both A-colocal objects and T A-colocal objects and C T A exists, then there is a natural isomorphism Proof. The assumption that T preserves A-colocal objects implies, by Theorem 4.7, that C A induces a colocalization C T on C T such that C A U ∼ = U C T naturally, and U preserves and reflects local objects and equivalences. Hence the C T -colocal objects are those (X, a) in 3 also tells us that there is no ambiguity in the right-hand term of (5.4), as it may indistinctly mean the underlying R-module of the localization of M with respect to S ⊗ R f in the category of left S-modules or the localization of the R-module underlying M with respect to the R-module homomorphism underlying S ⊗ R f , that is, Similarly, T preserves A-colocal objects for every left R-module A, and therefore we infer from Proposition 5.5 that there is a natural isomorphism for every left S-module M and every left R-module A. In fact, (5.5) really means that Both Proposition 5.3 and Proposition 5.5 have duals. If L f is a localization on C for some morphism f , then L f may be viewed as a colocalization on the opposite category C op whose colocal objects are those that are f -local in C, that is, those objects that are orthogonal to f in C and hence co-orthogonal to f in C op . Although it is actually the same functor L f , this colocalization could be denoted by C f for consistency. Similarly, if C A is a colocalization on C for some object A, then it is a localization on C op whose equivalences are those morphisms that are A-equivalences in C, and we denote this localization by L A . Accordingly, we call A-local those objects in a category that are A-colocal in the opposite category. In other words, the class of A-local objects in a category is the closure of A under double passage to the orthogonal complement. This notation and terminology is not of common use, so we confine it to the statement of the next result. The assumption that L A exists is really restrictive, since the opposite of a locally presentable category is not locally presentable. Corollary 5.7. Let U : C T ⇄ C : F be the Eilenberg-Moore factorization of a comonad T on a category C. Let A be an object in C such that L A exists. Then the following statements are equivalent: Moreover, if T preserves both A-local objects and T A-local objects and L T A exists, then there is a natural isomorphism L A U ∼ = L T A U . Homotopical localizations and cellularizations In the remaining sections we discuss localizations and colocalizations in a homotopical context, using the formalism of Quillen model categories [66], which we assume equipped with functorial factorizations. Every model category M can be endowed with homotopy function complexes as described in [52, § 17.5] is a natural weak equivalence of simplicial sets for X in M and Y in N , whose value at π 0 coincides with the bijection given by the derived adjunction (6.1). Unless otherwise specified, in the next sections we will consider this kind of simplicially enriched orthogonality between objects and maps in model categories. Thus, an object X and a map f : V → W in a model category M will be called simplicially orthogonal (or, as in [52, § 17.8], homotopy orthogonal ) if is a weak equivalence of simplicial sets. The objects that are simplicially orthogonal to every map in a given collection F are called F -local, and the maps that are simplicially orthogonal to the collection of all F -local objects are called F -equivalences. For convenience we omit in this article the standard assumption that F -local objects have to be fibrant (thus, our convention is that an object weakly equivalent to an F -local object is F -local). Every F -equivalence between F -local objects is a weak equivalence. If M satisfies suitable assumptions -for instance, if it is cofibrantly generated and left proper, and its underlying category is locally presentable, as for pointed or unpointed simplicial sets, Bousfield-Friedlander spectra [19], symmetric spectra over simplicial sets [54], groupoids [24], and many other cases-then for every set of maps F there exists a model category M F , called left Bousfield localization of M with respect to F , with the same underlying category as M and the same cofibrations, in which the weak equivalences are the F -equivalences and the fibrant objects are those that are fibrant in M and F -local; see [52, §3.3] for details. Left Bousfield localizations still exist if F is a proper class, provided that a suitable large-cardinal axiom holds [70]. An F -localization of an object X of M is a trivial cofibration l X : X → L F X in M F with L F X fibrant in M F . Thus if M F exists then L F is a fibrant replacement functor on M F . As such, (L F , l) is a monad on M that is idempotent on the homotopy category Ho(M). Therefore the classes of F -equivalences and F -local objects determine each other by ordinary orthogonality in Ho(M). In what follows, a homotopical localization on a model category M will mean an F -localization L F for some collection of maps F . This terminology is consistent with previous articles such as [23,26]. Although our results hold indeed for arbitrary collections of maps, we will state them for a single map f for simplicity of notation, since most motivating cases involve one map only. The next result is a homotopical version of Proposition 5.1. As explained in [52, § 3.1.11], we need to impose suitable fibrancy and cofibrancy assumptions due to the fact that F preserves weak equivalences between cofibrant objects but not all weak equivalences in general, and G preserves weak equivalences between fibrant objects. For a map f : V → W in M between cofibrant objects, the following assertions hold: such that α X • F l X = l F X , and for every object Y in N there is a homotopy unique and homotopy natural map where the vertical weak equivalences are given by (6.2) since F preserves cofibrant objects and G preserves fibrant ones, and the horizontal arrows are induced by F f and f . It follows that, if Y is fibrant, then GY is f -local if and only if Y is F f -local, as claimed in part (i). Therefore, G sends fibrant F f -local objects to f -local objects, and this implies that F sends f -equivalences between cofibrant objects to F f -equivalences. This proves (ii). Now (ii) implies that, for every cofibrant X in M, the map F l X is an F f -equivalence and hence a trivial cofibration in and α X is unique up to homotopy with this property. It also follows that α X is an F f -equivalence since F l X and l F X are F f -equivalences, and α X is a weak equivalence if and only if F L f X is F f -local (although it need not be fibrant). Naturality of α after passing to the homotopy categories is proved as in Theorem 3.2. Since the map l GY is a trivial cofibration in the model category M f and GL F f Y is fibrant in M f by part (i), there is a map It is unique up to homotopy since GL F f Y is f -local and l GY is an f -equivalence, and it is natural up to homotopy for similar reasons. Moreover, β Y is a weak equivalence if and only if it is an f -equivalence, which happens if and only if Gl Y is an f -equivalence. If F preserves weak equivalences then for every object X we may choose a cofibrant approximationX → X and we have FX ≃ F X, so the previous arguments hold withX in the place of X. Similarly, if G preserves weak equivalences, then for each Y we may choose a fibrant approximation Y →Ŷ and useŶ instead of Y . where sSet denotes the category of simplicial sets, and Proposition 6.1 yields a morphism for all X and all f , which is the natural πf -equivalence studied in [24]. From the fact that the morphism (6.4) is a πf -equivalence it follows that, if X is 1-connected, then L πf (πL f X) is trivial. It is a long-standing open problem to decide if L f X is in fact 1-connected for every map f when X is 1-connected. It should be possible to extend (6.4) to higher dimensions by using suitable categories of algebraic models for n-types, in which homotopical localizations can be effectively computed, as done in [24] for the model category of groupoids. This might yield relevant information on the n-type of L f X, which is usually difficult to relate with the n-type of X. Let us however emphasize that L πf (πX) is very different from the space L P1f (P 1 X), where P 1 denotes the first Postnikov section, that is, P 1 X = K(π 1 (X, x 0 ), 1). The spaces P 1 L f X and L P1f (P 1 X) are not P 1 f -equivalent in general. For example, let f be a map between wedges of circles such that L f X is the localization of X at the prime 3; thus where Σ 3 is the symmetric group on three letters, then, as shown in [27,Example 8.2], the space L f X = K(Σ 3 , 1) (3) is 1-connected since Σ 3 is generated by elements of order 2, yet it is not contractible since Σ 3 has nonzero mod 3 homology. Therefore, P 1 L f X is contractible while L P1f (P 1 X) = L f X is not. Although P 1 is left adjoint to the inclusion of the full subcategory of 1-coconnected spaces, the functor L P1f does not restrict to this subcategory; that is, L P1f (P 1 X) need not be a K(G, 1). If A is any collection of objects in a model category M, then a map g : X → Y will be called an A-equivalence if is a weak equivalence of simplicial sets for every A ∈ A. Correspondingly, an object B (for the purposes of this article, not necessarily cofibrant) is called A-colocal (or, more commonly, A-cellular ) if map M (B, g) is a weak equivalence for every A-equivalence g. As shown in [52, § 5.1], if suitable assumptions are imposed on M then there exists a model category M A for every set of objects A, called right Bousfield localization of M with respect to A, with the same underlying category as M and the same fibrations, in which the weak equivalences are the A-equivalences and the cofibrant objects are those that are cofibrant in M and A-colocal. In order to ensure the existence of M A it is sufficient to assume that M is right proper and cofibrantly generated [7, § 5]. In fact the latter condition can be weakened, as done in [30, § 2]. An A-colocalization (or A-cellularization) of an object X of M is a trivial fibration c X : Thus if M A exists then C A is a cofibrant replacement functor on M A . As such, (C A , c) is a comonad on M that is idempotent on the homotopy category Ho(M). A homotopical colocalization on a model category M will mean an A-cellularization C A for some collection of objects A. We will state our results for a single object A in the rest of the article for consistency with the examples, although most of our conclusions remain valid for collections of objects. weak equivalence if and only if GC F A Y is A-cellular, and β X is a weak equivalence if and only if F c X is an F A-equivalence. If F preserves all weak equivalences then the cofibrancy assumptions are not necessary, and if G preserves all weak equivalences then the fibrancy assumptions can be omitted. Proof. This is proved in the same way as Proposition 6.1. Remark 6.4. If an object C is B-cellular and B is A-cellular, then C is A-cellular. To check this, let g : X → Y be any A-equivalence. Then map M (B, g) is a weak equivalence. Hence g is a B-equivalence, and this implies that map M (C, g) is a weak equivalence, as needed. A similar argument in the case of localizations shows that if a map h is a g-equivalence and g is an f -equivalence, then h is an f -equivalence. (a) Let A be a cofibrant object of M. If T A is cofibrant and A-cellular, then the classes of F A-cellular objects and F T A-cellular objects coincide. (b) Let f be a map in M between cofibrant objects. If T f is an f -equivalence between cofibrant objects, then the classes of F f -equivalences and F T f -equivalences coincide. Proof. In part (a), since T A is cofibrant and A-cellular by assumption, we infer from part (ii) of Proposition 6.3 that F T A is F A-cellular. Conversely, F A is F T A-cellular since F is a retract of F U F = F T by (2.2). Hence, by Remark (6.4), the classes of F A-cellular objects and F T A-cellular objects are equal. Part (b) is proved with the same argument, using part (ii) of Proposition 6.1. We conclude this section with a result relating certain localizations with cellularizations, followed by examples that will be relevant in Section 9. In a pointed model category, localization with respect to a map A → * is called A-nullification and denoted by P A . A motivating example is A = S n+1 for n ≥ 0 in the category of pointed simplicial sets, for which P A is the nth Postnikov section. In this example, C A is the n-connected cover, so there is a homotopy fibre sequence for every space X if A = S n+1 . An analogous sequence exists for spectra. In fact, (6.5) is a homotopy fibre sequence of spectra if and only if a certain condition stated in [47, Theorem 3.6] holds, as is the case if A is any suspension of the sphere spectrum, and also whenever C A (and hence also P A ) commutes with suspension. The extent to which (6.5) fails to be a homotopy fibre sequence for spaces in general was discussed in [29]. Theorem 6.6. Let A be a cofibrant object in a pointed model category M such that C A and P A exist and for every object X the natural sequence is a homotopy fibre sequence. Let f : V → W be a map between A-cellular objects such that L f exists. Then there is a natural equivalence showing that C A Z is also f -local. Thus C A preserves f -local objects, so our result will follow by applying Corollary 3.7 within Ho(M) if we prove that C A preserves f -equivalences. With this purpose, let g : X → Y be an f -equivalence, and consider the commutative diagram Since P A V ≃ * and P A W ≃ * , the map f is a P A -equivalence. Therefore, by Remark 6.4, every f -equivalence is a P A -equivalence, so in particular P A (g) is a weak equivalence. Now, if Z is any f -local object, then (6.6) yields by [53, Corollary 6.4.2(c)] a commutative diagram whose rows are homotopy fibre sequences of simplicial sets: This diagram shows that C A (g) : C A X → C A Y is an f -equivalence, as needed. Example 6.7. In the category of pointed simplicial sets, if f is a map between connected spaces, then a space X is f -local if and only if its basepoint component X 0 is f -local. Since X 0 ≃ C A X for A = S 1 , Theorem 6.6 yields a natural equivalence L f X 0 ≃ (L f X) 0 for all X. More generally, if f is any map between n-connected spaces for n ≥ 0, and X n denotes the n-connected cover of a space X, then X is f -local if and only if X n is f -local, so Theorem 3.2 yields a natural map which is weak equivalence by Theorem 6.6. This fact was first found in [28,Theorem 5.2]. It does no longer hold if f is not a map between n-connected spaces; for example, it is well-known that localization with respect to K-theory lowers connectivity [60]. Theorem 6.6 also implies that, if f is any map between n-connected spectra, then L f X n ≃ (L f X) n for every spectrum X and all n ∈ Z. As a special case, for every map f of spaces and every spectrum X, where the superscript denotes the connective cover. By connective we mean (−1)-connected, that is, X c ≃ C S X, where S is the sphere spectrum. Preservation of algebras over monads in model categories In what follows we consider monads on model categories arising from Quillen adjunctions. For a monad T acting on a model category M, we denote by M T the category of T -algebras, as in the previous sections. A model structure on M T is called transferred or right-induced from the model structure on M if the forgetful functor U creates (that is, preserves and reflects) weak equivalences and fibrations in the Eilenberg-Moore factorization of T , An instance where a transferred model structure on algebras over a monad does not exist will be described in Example 8.4 below. If M T admits a transferred model structure, then the forgetful functor U is right Quillen and hence (7.1) is a Quillen adjunction. Therefore T preserves weak equivalences between cofibrant objects, since T = U F and F preserves weak equivalences between cofibrant objects while U preserves all weak equivalences. Moreover, there is a derived adjunction where Q is a cofibrant replacement functor on M, and fibrant replacement on M T is omitted since U preserves all weak equivalences. The adjunction (7.2) is not equivalent, in general, to the Eilenberg-Moore adjunction of the derived monad U F Q = T Q on Ho(M). From now on, for simplicity, we will omit the cofibrant replacement functor Q from the notation, thus writing Ho(M) T instead of Ho(M) T Q . Counterexamples showing that the categories Ho(M T ) and Ho(M) T need not be equivalent can be found in [44] and [56]. In order to emphasize the distinction, we will call T -algebras up to homotopy the objects of Ho(M) T , while we will keep calling T -algebras those of Ho(M T ). The question of when an f -localization or an A-cellularization on M induces respectively a localization or a colocalization on T -algebras up to homotopy can be quickly answered using results from Section 4, as follows. in M to f -equivalences. Indeed, if g : X → Y is an f -equivalence then so is the map Qg : QX → QY , and if T sends f -equivalences between cofibrant objects to f -equivalences then T Qg is an f -equivalence. Conversely, suppose that T Q preserves f -equivalences. If g : X → Y is an f -equivalence where X and Y are cofibrant, then T Qg is an f -equivalence by assumption. Moreover T QX ≃ T X and T QY ≃ T Y since T preserves weak equivalences between cofibrant objects, so T g is an f -equivalence. In the case of A-cellularizations we use Theorem 4.7 to infer that C A lifts to Ho(M) T if and only if T Q preserves A-cellular objects in Ho(M). The assertion (b) follows since T QX ≃ T X for every cofibrant object X. We next address the existence of liftings of f -localizations to the homotopy category of T -algebras Ho(M T ). This lifting problem and the corresponding one for cellularizations have also been discussed in [9,76,77,78,79] and in [46,47,48,49]. We start by formalizing the notion that a homotopical localization or a homotopical colocalization on M lift to Ho(M T ). As next shown, it follows automatically that (L T , l T ) is then a localization and, correspondingly, (C T , c T ) is a colocalization. Proof. In order to prove that (L T , l T ) is an idempotent monad, it suffices to check that l T L T and L T l T are isomorphisms on Ho(M T ). On one hand, hL T • lU L T = U l T L T . Moreover, Lh • lLU = lU L T • h and lL is an isomorphism in Ho(M), so U l T L T is also an isomorphism. Since U reflects isomorphisms, l T L T is an isomorphism. On the other hand, by the naturality of h, we have hL T • LU l T = U L T l T • h. Here LU l T = Lh • LlU , and, since h and Ll are isomorphisms, it follows that LU l T is an isomorphism and hence so is U L T l T . Again, since U reflects isomorphisms, we conclude that L T l T is an isomorphism. The fact that U preserves and reflects local objects and equivalences follows from the isomorphism LU ∼ = U L T as detailed in Corollary 3.4, and the argument for colocalizations is analogous. In what follows we treat first the case of colocalizations for simplicity. For an object A in a model category M, we indistinctly denote by C A the A-cellularization functor on M and the induced colocalization on Ho(M), although it is important not to confuse them. (i) C A lifts to Ho(M T ). (ii) U sends F A-cellular objects to A-cellular objects. Proof. If (C A , c) lifts to Ho(M T ), then there is a colocalization functor with a natural transformation c T : C T → Id and a natural isomorphism h : C A U → U C T in Ho(M) such that U c T • h = cU . By Lemma 7.3, a map g in M T is a C T -equivalence if and only if U g is an A-equivalence, and part (i) of Proposition 6.3 tells us that U g is an A-equivalence if and only if g is an F A-equivalence. Hence C T and C F A are colocalizations with the same class of equivalences, so there is a natural isomorphism g : On the other hand, according to Proposition 6.3, for every T -algebra Y there is a homotopy unique and homotopy natural A-equivalence where we need not assume that Y is fibrant since U preserves all weak equivalences. By its uniqueness up to homotopy, α Y coincides with U g Y • h Y in Ho(M) and therefore α Y is an isomorphism for every T -algebra Y , thus yielding (iii). We next prove that (ii) and (iii) are equivalent. Indeed, α Y is a weak equivalence if and only if U C F A Y is A-cellular. Hence α Y is a weak equivalence for all Y if and only if U sends F A-cellular objects to A-cellular objects, as claimed. As one might have expected, the necessary and sufficient condition (ii) for the existence of a lifting of a cellularization C A to Ho(M T ) given in Theorem 7.4 is stronger than the one given in Theorem 7.1 for the existence of a lifting of C A to Ho(M) T . Proof. If C A lifts to Ho(M T ) then, by Theorem 7.4, U sends F A-cellular objects to A-cellular objects. According to part (ii) of Proposition 6.3, F sends cofibrant A-cellular objects to F A-cellular objects. Therefore, T = U F sends cofibrant A-cellular objects to A-cellular objects, and consequently C A lifts to Ho(M) T by part (b) of Theorem 7.1. Corollary 7.6. If A and T A are cofibrant and both C A and C T A lift to Ho(M T ), then there is a natural isomorphism Proof. According to Theorem 7.4, our assumptions yield natural isomorphisms in Ho(M) for all T -algebras Y . The fact that C A lifts to Ho(M T ) implies, by Theorem 7.4, that U sends F A-cellular objects to A-cellular objects. Hence T A = U F A is A-cellular, and Proposition 6.5 implies then that the classes of F A-cellular objects and F T A-cellular objects coincide. This means that C F A ∼ = C F T A , hence completing the argument. A favorable situation for the use of Corollary 7.6 is when the forgetful functor U has a right adjoint. An important case when this happens will be discussed in the next section. Proposition 7.7. Suppose that the forgetful functor U : M T → M has a Quillen right adjoint. If T A is A-cellular, then C A and C T A lift to Ho(M T ) and there is a natural Proof. Since both F and U are Quillen left adjoints and A is assumed to be cofibrant, F A and T A = U F A are cofibrant. Moreover, the fact that U is a Quillen left adjoint implies, by part (ii) of Proposition 6.3, that U sends cofibrant F A-cellular objects to T A-cellular objects, and, since U preserves weak equivalences, U sends all F A-cellular objects to T A-cellular ones. Now the assumption that T A is A-cellular implies, by Remark 6.4, that T A-cellular objects are A-cellular and therefore U sends F A-cellular objects to A-cellular objects. Therefore C A lifts to Ho(M T ) by Theorem 7.4. Moreover, according to part (a) of Proposition 6.5, the classes of F A-cellular objects and F T A-cellular objects coincide. Hence U sends F T A-cellular objects to T A-cellular objects, and this implies that C T A also lifts to Ho(M T ), so Corollary 7.6 applies. We next discuss the case of homotopical localizations, which is largely similar. Proof. Recall from Proposition 6.1 that for every T -algebra Y there is a homotopy unique and homotopy natural map Hence β Y is a weak equivalence for all Y if and only if U sends all F f -localization maps to f -equivalences, and this happens if and only if U sends F f -equivalences to f -equivalences. This proves that (ii) ⇔ (iii). The claim (i) means that there is a localization functor with a natural transformation l T : Id → L T and a natural isomorphism h : Hence L T and L F f are localizations with the same essential image class, so there is a natural isomorphism g : L T → L F f in Ho(M T ) with g • l T = l. By its uniqueness up to homotopy, β Y coincides with U g Y • h Y in Ho(M) and hence β Y is an isomorphism for every T -algebra Y , which proves (iii). By definition (iii) implies (iv), and (iv) implies (i). Proof. The domain and codomain of T f are cofibrant since U preserves cofibrations and hence so does T . By part (b) of Proposition 6.5, the class of F T f -equivalences coincides with the class of F f -equivalences. Hence, in order to infer that L T f lifts to Ho(M T ), we need to prove that U sends F f -equivalences between cofibrant objects to T f -equivalences. This follows from part (ii) of Proposition 6.1 since U is, by assumption, a Quillen left adjoint. Hence our claim is implied by Corollary 7.11. There is, however, a relevant distinction between right Bousfield localizations and left Bousfield localizations of categories of T -algebras equipped with transferred model structures, due to the fact that in a right-induced model structure one asks the forgetful functor to create fibrations but not cofibrations. Indeed, as shown in Theorem 7.13 below, the equality of model categories always holds, while the analogous equality for f -localizations is not necessarily true, and in fact if it holds then L f lifts to T -algebras. The ideas behind the next theorem were found by Gutiérrez-Röndigs-Spitzweck-Østvaer in [48] for algebras over coloured operads, and also by Batanin-White in [9, Theorem 3.4] and White-Yau in [79, Corollary 2.8] by weakening the assumption that M T has a transferred model structure. hence F preserves cofibrations and U preserves and reflects trivial fibrations. Moreover, part (i) of Proposition 6.1 says that U preserves and reflects fibrant objects, while part (ii) tells us that F preserves weak equivalences between cofibrant objects. If a transferred model structure exists for (M f ) T , then it has the same trivial fibrations and the same fibrant objects as (M T ) F f , since these are created by U in both cases. Hence they also have the same cofibrations -and thus the same cofibrant objects-and the same weak equivalences between cofibrant objects, since these are determined by the fibrant objects. Therefore they have the same weak equivalences (since cofibrant approximations are trivial fibrations) and this proves that, indeed, (M f ) T = (M T ) F f as model categories. From this fact it follows that U : M T → M sends F f -equivalences to f -equivalences, which, according to Theorem 7.8, implies that L f lifts to Ho(M T ). Consequently, in order to display a case where a transferred model structure for (M f ) T does not exist, it is enough to exhibit an f -localization that does not lift to T -algebras. This is done at the end of the next section. Module spectra The category Sp of symmetric spectra over simplicial sets has a closed symmetric monoidal structure with internal function spectrum Hom S (−, −) defined as in [54,Definition 2.2.9]. We consider the stable model structure [54,Theorem 3.4.4] on Sp, which, according to [54,Theorem 4.2.5], is Quillen equivalent to the Bousfield-Friedlander model structure on the category of ordinary spectra. If Q is a cofibrant replacement functor and R is a fibrant replacement functor on Sp, we call F(X, Y ) = Hom S (QX, RY ) a derived function spectrum. At the same time, Sp is a simplicial category with enrichment defined as for all n, where ∆[n] + denotes the standard n-simplex with a disjoint basepoint. Thus, Map(QX, RY ) is a convenient choice of a homotopy function complex from X to Y in Sp. Moreover, it follows from [53, Lemma 6.1.2] or [54, Corollary 2.2.11] that the homotopy groups of the simplicial set Map(QX, RY ) are isomorphic to those of the spectrum F(X, Y ) in non-negative dimensions, hence to those of the connective cover F c (X, Y ) in all dimensions. As a consequence of this fact, homotopy orthogonality (6.3) in Sp can be formulated in terms of F c (−, −), as done in [16,18,25,46] and in many other articles. For any ring spectrum E, the category E-Mod of left E-module spectra admits, by [ We say that a homotopy functor on spectra is triangulated if it preserves cofibre sequences. For a localization or a cellularization, this condition is equivalent to commuting with suspension, as shown in [25,Theorem 2.7] and [47,Theorem 2.9]. Localizations and cellularizations need not be triangulated. However, C A is triangulated if A = ∞ i=0 Σ −i W for some spectrum W -this condition was also considered in [5,Lemma 5.7]-, and L f is triangulated if f = ∞ i=0 Σ i g for some map g. Localizations with respect to homology theories [14] are triangulated. Theorem 8.1. Let A and E be cofibrant symmetric spectra and suppose that E is a ring spectrum. Let U be the forgetful functor from left E-modules to spectra. If E is connective or C A is triangulated, then C A lifts to left E-modules and Proof. Since U is a Quillen left adjoint, we are ready to apply Proposition 7.7, and it will suffice to show that U (E ∧A) is A-cellular. For this we need to assume, as in [45, § 4] and [47,Theorem 4.1], that E is connective or C A is triangulated. If E is connective, then we may use the connective cover F c (−, −) of the derived function spectrum instead of a homotopy function complex to infer that every A-equivalence g : X → Y induces . If E is not necessarily connective but C A is triangulated, then each A-equivalence g : X → Y induces a weak equivalence F(A, X) ≃ F(A, Y ) by [47, Theorem 2.9], and therefore as needed. In conclusion, if E is connective or C A is triangulated, then C A lifts to Ho(E-Mod) and Proposition 7.7 tells us that C A U Y ∼ = C U(E∧A) U Y for every cofibrant spectrum A and every left E-module Y . Moreover Theorem 7.4 implies that C E∧A is a lift of C A and hence C A U Y ≃ U C E∧A Y for every left E-module Y , as claimed. Example 8.2. As an easy example, pick A = S (the sphere spectrum), so that the S-cellular spectra are precisely the connective ones. It then follows from Theorem 8.1 that, if E is a connective ring spectrum, then C E X ≃ C S X ≃ X c for every spectrum X underlying a left E-module. Hence, connective E-modules are E-cellular. Theorem 8.3. Let E be a cofibrant symmetric spectrum and let f be a map between cofibrant spectra. Suppose that E is a ring spectrum, and let U be the forgetful functor from left E-modules to spectra. If E is connective or L f is triangulated, then L f lifts to left E-modules and Proof. As in Theorem 8.1, we use the fact that the forgetful functor U has a Quillen right adjoint, namely Hom S (E, −). In view of Proposition 7.12, it is enough to prove that U (E∧f ) is an f -equivalence. This happens provided that E is connective or L f is triangulated, by a similar argument as in the proof of Theorem 8.1; see also [25,Theorem 2.7]. Hence, under these assumptions, L f lifts to Ho(E-Mod) and Proposition 7.12 together with Theorem 7.8 imply that Example 8.4. Let K(n) be the nth Morava K-theory spectrum at a prime p for any n ≥ 1, and let k(n) be its connective cover. Then HZ/p ∧ K(n) = 0 while HZ/p ∧ k(n) = 0, as proved in [67,Theorem 2.1]. This implies that k(n) is not a homotopy retract of K(n)∧k(n) and therefore k(n) cannot be a left K(n)-module. Here k(n) ≃ C S K(n) where S denotes the sphere spectrum; consequently, C S does not lift to K(n)-modules. This counterexample shows that a cellularization C A of spectra need not lift to E-modules if the assumptions that E is connective or C A is triangulated in Theorem 8.1 both fail to hold. Note that C S is not triangulated since it converts the cofibre sequence Similarly, if one removes the assumption that E is connective or L f is triangulated from Theorem 8.3, then it need no longer be true that L f lifts to E-module spectra. Indeed, the cofibre of the canonical map k(n) → K(n) is the Postnikov section P −1 K(n); since HZ/p ∧ P −1 K(n) = 0, we find that P −1 K(n) is not a homotopy retract of K(n) ∧ P −1 K(n) and this implies that P −1 K(n) cannot be a left K(n)-module. But P −1 K(n) = L f K(n) with f : S → 0. Thus, if f : S → 0 and E = K(n), then L f does not lift to E-modules. Example 8.4 also shows that a transferred model structure for (M f ) T need not exist for a monad T acting on a model category M. This is a consequence of part (b) of Theorem 7.13, since in Example 8.4 we have exhibited a case where L f does not lift to Ho(M T ). Loop spaces and infinite loop spaces In this section we consider simplicial operads (that is, operads taking values in simplicial sets) acting on pointed simplicial sets endowed with the Cartesian product. This choice is due to the fact that important monads such as the James construction, the Q-construction, or the infinite symmetric product, which will be used in this section, involve products and basepoints in their definition. Another reason is that in the unpointed category cellularizations are trivial due to the fact that every space X is a retract of map(A, X) if A is nonempty, as pointed out in [33, Remark 2.A.1.1] and [52,Remark 3.1.10]. In the case of localizations, there are no essential differences between working with basepoints or without them. In other words, map * (−, −) or map(−, −) can indistinctly be used to test homotopy orthogonality for localizations of pointed connected spaces [17,Lemma 2.1], and hence if a map g of pointed spaces is an f -equivalence then both g × K and g ∧ K are f -equivalences for every pointed simplicial set K. However, if a pointed space X is A-cellular then X ∧ K is A-cellular for all K but X × K need not be (since it has K as a retract). For a simplicial operad P and every n, the (unpointed) simplicial set P (n) is equipped with an action of the symmetric group Σ n . We assume, as usual, the existence of a unit element u ∈ P (1), and we will also assume that operads are reduced, meaning that P (0) is a single point, denoted by * . If P is such an operad and X is a simplicial set with a basepoint x 0 then a P -algebra structure on X is a morphism of operads P → End(X) where End(X)(n) is the based function complex map * (X n , X), as in [59, § 1]. Here we denote by X n the n-fold Cartesian product of X with itself, which is meant to be a point if n = 0, and pick (x 0 , . . . , x 0 ) as basepoint of X n if n = 0. The operad End(X) is reduced and the image of the unit element u ∈ P (1) under the structure map P (1) → map * (X, X) is assumed to be the identity map. For a simplicial operad P acting on pointed simplicial sets through basepoint-preserving maps, as explained in [59], the P -algebras coincide with the algebras over the reduced monad T where T X is defined, for a pointed simplicial set X, as the quotient of (9.1) n≥0 P (n) × Σn X n by identifiying (w, s i y) ∈ P (n) × X n with (σ i w, y) ∈ P (n − 1) × X n−1 for 1 ≤ i ≤ n and all w ∈ P (n) and y ∈ X n−1 . Here the map s i : X n−1 → X n inserts the basepoint into the i-th place and σ i : P (n) → P (n − 1) is the i-th degeneracy, defined as σ i w = γ i (w, u, . . . , u, * , u, . . . , u), is among the multiplication maps of the operad P ; see [59, § 4]. Thus, in particular, P (0) is identified with (w, x 0 , . . . , x 0 ) ∈ P (n) × X n for every n and all w ∈ P (n). The unit map η X : X → T X of the monad sends each element x to (u, x) ∈ P (1) × X. The same happens for non-symmetric operads by discarding the Σ n -action from (9.1). For example, if A is the (non-symmetric) unital associative operad, for which A(n) is a single point for all n, then the associated reduced monad on pointed simplicial sets is the James construction [55]. Lemma 9.1. If T is the reduced monad associated with a simplicial operad P acting on pointed simplicial sets, then T preserves f -equivalences for every map f . Proof. For a pointed simplicial set X, we write T X as a homotopy colimit of the partial sums T k X = 0≤n≤k P (n) × Σn X n subject to the same identifications as in (9.1). Thus, T 0 X = P (0), T 1 X ∼ = P (1) + ∧ X, and T k X is a pointed simplicial set such that where X ∧k denotes X ∧ · · · ∧ X with k factors. Hence for each k and every pointed simplicial set Z there is a Kan fibre sequence Since the smash product of two f -equivalences is an f -equivalence, if g : X → Y is an f -equivalence then so is the induced map for every finite number of factors and every pointed simplicial set W , where (9.3) is the identity on the first factor and g on the other factors. If the operad P is non-symmetric then we can omit the Σ n -action and (9.2) proves inductively that if g : X → Y is an f -equivalence then T k g is an f -equivalence for every k and consequently T g is also an f -equivalence. If P is symmetric, we need to use the fact that the quotient P (k) + ∧ Σ k X ∧k is a colimit over Σ k (viewed as a small category), and it is also a homotopy colimit if Σ k acts freely on P (k), but not otherwise. However, P (k) + ∧ Σ k X ∧k is a homotopy colimit of a (free) diagram indexed by the opposite of the orbit category of Σ k , where the value of the diagram at Σ k /G is the fixed-point space P (k) G + ∧ (X ∧k ) G ; cf. [33, § 4.A.4]. Since each space (X ∧k ) G is homeomorphic to a product X ∧m with m ≤ k, we obtain that the map T g : T X → T Y is a homotopy colimit of f -equivalences, and therefore it is itself an f -equivalence. Essentially the same argument works for cellularizations. We need the assumption that A be connected in order to avoid triviality, since S 0 is a retract of every non-connected simplicial set and all simplicial sets are S 0 -cellular. Lemma 9.2. If T is the reduced monad associated with a simplicial operad P acting on pointed simplicial sets and A is connected, then T preserves A-cellular simplicial sets. Proof. If X is A-cellular, then every finite smash product P (k) + ∧ X ∧k is A-cellular for k ≥ 1 by [33, Theorem 2.D.8], while P (0) = * by assumption, which is also A-cellular. The proof continues similarly as in Lemma 9.1. Example 9.3. The infinite symmetric product [32], denoted by SP ∞ , is the reduced monad associated with the commutative operad. Its algebras are the commutative monoids. The argument given in the proof of Lemma 9.1 was used in [28,Theorem 1.3] to prove that SP ∞ preserves f -equivalences for every map f . Moreover, it was shown in [28, Proposition 1.1] that a space X underlies an SP ∞ -algebra in the pointed homotopy category if and only if X is a GEM, i.e., a weak product of abelian Eilenberg-Mac Lane spaces; moreover, in this case the SP ∞ -algebra structure on X is unique up to isomorphism. Therefore, by Lemma 9.1 and part (a) of Theorem 7.1, every f -localization preserves GEMs (as first shown in [33,Chapter 4]) and defines in fact a localization on the homotopy category of GEMs. Similarly, every A-cellularization preserves GEMs and defines a colocalization on them by Lemma 9.2 and part (b) of Theorem 7.1. Remark 9.4. Lemma 9.1 also holds for the unreduced monadT associated with a simplicial operad P , that is,T X = n≥0 P (n) × Σn X n without any basepoint identifications, whose algebras are also the P -algebras [59, § 4]. The proof proceeds with the same argument as in the proof of Lemma 9.1, using the fact that if X → Y is an f -equivalence then W × X n → W × Y n is also an f -equivalence for every simplicial set W and all n. However, Lemma 9.2 is not true for unreduced monads, since, for an A-cellular space X, a product W × X is not A-cellular unless W is itself A-cellular. Still, Lemma 9.2 holds for unreduced monads if we assume that each P (n) and all fixedpoint spaces P (n) G are contractible or empty for every subgroup G ⊆ Σ n . This condition is trivially satisfied if P (n) is a single point for all n and also if each P (n) is contractible and Σ n acts freely on it. This is the case, for instance, for the commutative operad and its cofibrant approximations. We next address liftings of localizations L f and cellularizations C A to categories of algebras over simplicial operads. The existence of a transferred model structure for such categories of algebras is guaranteed by results in [12]. In order to apply Theorem 7.4 and Theorem 7.8 to the associated monad T = U F , we need to prove that the forgetful functor U sends F f -equivalences of T -algebras to f -equivalences of spaces for every basepointpreserving map f , and that U sends F A-cellular T -algebras to A-cellular spaces for every pointed connected space A. Both statements would be straightforward if U preserved homotopy colimits. However, in general, U only preserves sifted colimits (including filtered ones and reflexive coequalizers), since T commutes with these due to the fact that sifted colimits commute with finite products of simplicial sets; cf. [56,Proposition 2.5]. To surmount this difficulty, we rely on a method used by Gutiérrez-Röndigs-Spitzweck-Østvaer in [48, § 3] with a similar purpose. For a T -algebra X, we consider the standard simplicial resolution of X by free T -algebras (compare with May's two-sided bar construction [58, § 9]), defined as for n ≥ 0. Thus B * X is a simplicial T -algebra with face and degeneracy maps coming from the unit and the counit of the adjunction between F and U . Moreover, the counit F U X → X yields a map B * X → X of simplicial T -algebras, where X is treated as a constant simplicial T -algebra. The monad T commutes with geometric realization since T is defined by means of a coend formula, namely (9.1), and geometric realization is itself a coend. It then follows that the forgetful functor U also commutes with geometric realization by the argument given in [56,Proposition 3.12]. This implies that the map B * X → X induces a weak equivalence \|B * X\| ≃ X of T -algebras, since U reflects weak equivalences and the map U \|B * X\| → U X has a left homotopy inverse induced by the unit U X → U F U X; see [56,Proposition 3.13]. Lemma 9.5. Let T be the reduced monad associated with a simplicial operad acting on pointed simplicial sets, and let T = U F be its Eilenberg-Moore factorization. (a) Let f be a basepoint-preserving map and let G be a collection of maps of T -algebras such that U g is an f -equivalence for every g ∈ G. Then U h is an f -equivalence for every h in the closure of G under pointed homotopy colimits. (b) Let A be a pointed connected simplicial set, and let D be a collection of T -algebras such that U D is A-cellular for every D ∈ D. Then U X is A-cellular for every X in the closure of D under pointed homotopy colimits. Proof. We first prove (b). Suppose inductively that X ≃ hocolim i∈I D i where I is a small category and U D i is A-cellular for every i. If we denote J n = hocolim i∈I B n (D i ), then, since geometric realization commutes with colimits, Moreover, since U preserves weak equivalences and commutes with geometric realization, we find that U X ≃ U \|J * \| ≃ \|U J * \|. Hence, in order to infer that U X is A-cellular, it is enough to prove that U J n is A-cellular for all n. Indeed, as F commutes with colimits, Since U D i is A-cellular for every i and T preserves A-cellular simplicial sets by Lemma 9.2, we conclude that U J n is A-cellular, as needed. This proves part (b). The proof of part (a) follows the same argument in the category of maps of T -algebras, using that T preserves f -equivalences by Lemma 9.1. To avoid ambiguity with "pushouts of maps", note that the left-hand square / / g is a pushout square in a category C if and only if the right-hand square is a pushout square in the category of arrows of C. Theorem 9.6. Let T be the reduced monad associated with a simplicial operad P acting on pointed simplicial sets. (a) For every basepoint-preserving map f : V → W , the functor L f lifts to Ho(sSet T * ). (b) For every pointed connected simplicial set A the functor C A lifts to Ho(sSet T * ). Proof. Let T = U F be the Eilenberg-Moore factorization of T . We aim to apply Theorem 7.4 and Theorem 7.8. The category sSet T * of T -algebras has a transferred model structure by [12, Proposition 4.1(c)], which is cofibrantly generated by [31] and locally presentable by [1, § 2.78], since T preserves filtered colimits. It is right proper as U preserves limits, fibrations, and weak equivalences, and consequently right Bousfield localizations (sSet T * ) F A exist for all A. It is not known to the authors if left properness of sSet T * holds in general, although it does if the given operad P is cofibrant, according to [75,Theorem 4.3], and also under the assumptions made in [63, Definition 3.1], which hold for the unital associative operad among others. However, for our purposes it will be enough that the model category sSet T * be Quillen equivalent to a left proper combinatorial model category, and this is guaranteed by [35,Corollary 1.2]. This fact ensures that, for every map f , a localization exists on Ho(sSet T * ) whose equivalences are the F f -equivalences, and this suffices to provide a lifting of L f to Ho(sSet T * ) if the remaining assumptions in Theorem 7.8 are fulfilled (see Remark 7.9). Thus, for part (a) we need to prove that the forgetful functor U sends F f -equivalences of T -algebras to f -equivalences of spaces, and for part (b) we want to see that U sends F A-cellular T -algebras to A-cellular spaces. We treat the latter first. The class of F A-cellular T -algebras is the smallest class of T -algebras containing F A and closed under pointed homotopy colimits [52, § 5.5]. Since U F A = T A is A-cellular by Lemma 9.2, the result follows from part (b) of Lemma 9.5 with D = {F A}. As for (a), we note that, since T preserves f -equivalences by Lemma 9.1, the F f -equivalences coincide with the F T f -equivalences by Proposition 6.5. What we will next prove, for technical convenience, is that U sends F T f -equivalences to f -equivalences. In fact it is enough to prove that U sends all localization maps l X : Each localization map l X is constructed as a possibly transfinite composite of homotopy pushouts of generating trivial cofibrations in sSet T * and horns on F T f with n ≥ 0: where we are using the simplicial structure of sSet T * described in [56, Proposition 2.14]. Specifically, for a T -algebra X with structure map α : T U X → U X and a simplicial set K, the T -algebra X ⊗ K is defined by means of a reflexive coequalizer: where β is adjunct to the map T U X × K → T (U X × K) given by the fact that T is a simplicial functor, and the common section is is an f -equivalence for every K, since T preserves f -equivalences by Lemma 9.1. Then Lemma 9.5 tells us that U h is an f -equivalence for every map h that can be constructed from maps of the form F T f ⊗ K by means of homotopy pushouts. This implies that U (λ n,f ) is an f -equivalence for every horn λ n,f , since λ n, . Finally, as l X can be constructed from horns and weak equivalences by means of homotopy pushouts, we conclude that U l X is an f -equivalence using Lemma 9.5 again. In what follows, Ω denotes the derived loop functor on pointed simplicial sets, that is, ΩX = map * (S 1 , RX), where S 1 = ∆[1]/∂∆ [1] and R is a fibrant replacement functor. Let A be the (non-symmetric) unital associative operad, for which A(n) is a single point for all n, and let ϕ : A ∞ → A be a cofibrant resolution [61,62]. If we let A act on pointed simplicial sets then the corresponding algebras are the monoids and the associated reduced monad is the James construction [55]. The morphism of operads ϕ : A ∞ → A yields a Quillen equivalence so the homotopy category of monoids is equivalent to the homotopy category of A ∞ -algebras in pointed simplicial sets. Moreover, the classifying space functor B is part of an adjunction (9.5) B : Ho(A ∞ -alg) ⇄ Ho(sSet * ) : Ω, which restricts, as a special case of (2.3), to an equivalence of categories between the full subcategory of Ho(A ∞ -alg) whose objects are those M such that the unit η M : M → ΩBM is an isomorphism (that is, grouplike spaces) and the full subcategory of connected simplicial sets, which are precisely those X for which the counit ε X : BΩX → X is an isomorphism. Proof. Let F : sSet * ⇄ A-alg : U be the Eilenberg-Moore factorization of the James construction J as a monad on pointed simplicial sets. Theorem 9.6 tells us that L f lifts to Ho(A-alg), and we then infer from the equivalence of (i) and (iii) in Theorem 7.8 that L f U M ≃ U L F f M for every monoid M . Let X be any pointed simplicial set. Since ϕ : A ∞ → A allows rectification of algebras, there is a monoid M X whose underlying space is weakly equivalent to ΩX. Hence, Next we observe that, since homotopical localizations preserve π 0 by Example 6.7, the functor L F f restricts to the full subcategory of Ho(A ∞ -alg) whose objects are grouplike and L BF f restricts to the full subcategory of connected simplicial sets. Since (9.5) sets up an equivalence between these two categories, B preserves both local objects and equivalences and, by Theorem 3.2, there is a comparison map of β type which is a weak equivalence. Now BF f ≃ Σf since f is a map of connected spaces and JY ≃ ΩΣY if Y is connected. Moreover, if X 0 denotes the basepoint component of X, then ΩL Σf BM X ≃ ΩL Σf X 0 ≃ ΩL Σf X, and this yields (i). Finally, it follows from Corollary 7.11 that as claimed in part (ii). By induction we also have L f Ω n X ≃ Ω n L Σ n f X for all X and n ≥ 0. This formula also holds for n = ∞, by the following argument, which is similar to the preceding one. Let E be the commutative operad, for which E(n) is a point for all n, and let ψ : E ∞ → E be a cofibrant resolution. The algebras over E in pointed simplicial sets are the commutative monoids and the associated reduced monad is the infinite symmetric product SP ∞ . Connected commutative monoids are GEMs. For a space X, the quotient X n /Σ n by the symmetric group action does not have the same homotopy type for n ≥ 2 as C(n) × Σn X n , where C(n) is a contractible space with a free Σ n -action. For this reason, E ∞ -spaces are not homotopy equivalent to commutative monoids, in general. Instead, if we denote by B ∞ X the Ω-spectrum associated with a given E ∞ -space X, then there is an adjunction which restricts, as another instance of (2.3), to an equivalence of categories between the full subcategory of grouplike E ∞ -spaces (i.e., infinite loop spaces) and the full subcategory of connective spectra; see [3,Pretheorem 2.3.2]. If Q denotes the reduced monad associated with E ∞ on pointed simplicial sets, then May's Approximation Theorem [58] implies that QX ≃ Ω ∞ Σ ∞ X if X is connected. Corollary 9.8. If f is any basepoint-preserving map between connected simplicial sets, then Proof. Let F : sSet * ⇄ E ∞ -alg : U be the Eilenberg-Moore factorization of the reduced monad Q associated with E ∞ . The functor L f lifts to Ho(E ∞ -alg) by Theorem 9.6, and it follows from Theorem 7.8 that For a spectrum X, we may view Ω ∞ X as an E ∞ -algebra. Hence, Since homotopical localizations preserve π 0 , the functor L F f restricts to the full subcategory of Ho(E ∞ -alg) whose objects are grouplike, and, since is connected, the functor L B ∞ F f restricts to the full subcategory of connective spectra as explained in Example 6.7. Hence Theorem 3.2 yields a comparison map of β type which is a weak equivalence. Here B ∞ F f ≃ Σ ∞ f and, if X c is the connective cover of X, where the last step uses (6.8). This yields (i), and it follows from Corollary 7.11 that as claimed in part (ii). Similarly, for cellularizations we have the following. Corollary 9.9. If A is any pointed connected simplicial set, then (i) C A ΩX ≃ ΩC ΣA X, and (ii) C A ΩX ≃ C ΩΣA ΩX for all pointed simplicial sets X, and Proof. The proof follows the same steps as the proofs of Corollary 9.7 and Corollary 9.8, using part (b) of Theorem 9.6. Some of the formulas obtained in the preceding results are contained in [15,16,33]. The preservation of loop spaces (and, more generally, spaces with an action of an algebraic theory) by homotopical localizations was also addressed by Badzioch in [4]. 10. Algebras up to homotopy Suppose given a monad T on a model category M. In this section we will not consider model structures on the category M T of T -algebras, but we will assume that T preserves weak equivalences and hence descends to a monad on the homotopy category Ho(M). As in Section 7, objects of the Eilenberg-Moore category Ho(M) T will be called T -algebras up to homotopy. Our aim in this section is to prove that the equivalence L f X ≃ L T f X obtained in Corollary 7.11 also holds when X underlies a T -algebra up to homotopy, provided that T and f interact in a suitable way, and similarly for cellularizations. For this, it will be convenient to work with the Dwyer-Kan construction of homotopy function complexes as simplicial sets of morphisms in the hammock localization of M, as in [39,40] or in the more elaborate version discussed in [38, § 35.6]. Thus, in this section we choose map M (X, Y ) to be the colimit of the nerves N L n (X, Y ), where L n (X, Y ) is the category whose objects are strings of n maps in M in arbitrary directions (10.1) X = X 0 ←→ X 1 ←→ X 2 ←→ · · · ←→ X n−1 ←→ X n = Y, where the arrows pointing backwards are weak equivalences. A morphism in L n (X, Y ) is a commuting diagram between strings of the same type. The colimit of nerves is taken along the maps induced by the functors L n (X, Y ) → L n+1 (X, Y ) consisting of adding id Y at the end of (10.1). Choosing instead to add id X or interpolating identities at any other place would replace map M (X, Y ) by a weakly equivalent simplicial set, since natural transformations of functors yield simplicial homotopies between maps after taking nerves. Suppose given a T f -local object X together with a map a : T X → X such that a • η X ≃ id X . Then X is f -local. Proof. By assumption, map M (T f, X) is a weak equivalence. Hence, we will achieve our goal if we prove that map M (f, X) is a homotopy retract of map M (T f, X) and therefore it is a weak equivalence as well. For this, consider the diagram Since T preserves f -equivalences between cofibrant objects, it follows from Theorem 4.2 that there is a localization L ′ on Ho(M) T such that U preserves and reflects local objects and equivalences, and L f U X ≃ U L ′ X naturally for every homotopy T -algebra X. Since T also preserves T f -equivalences, we infer similarly that L T f U X ≃ U L ′′ X for a localization L ′′ . We next show that L ′ ≃ L ′′ as in the proof of Proposition 5.3. The L ′ -local objects are those (X, a) in Ho(M) T such that X is f -local, and the L ′′ -local objects are those such that X is T f -local. Since T preserves f -equivalences between cofibrant objects, T f is an f -equivalence and hence every f -local object of M is T f -local. This tells us that all L ′ -local objects of Ho(M) T are L ′′ -local. Conversely, if (X, a) is L ′′ -local in Ho(M) T , then we may assume that X is fibrant and cofibrant and T f -local. Since T preserves cofibrant objects, we may also assume that a : T X → X is represented by a map in M. Thus, a • η X ≃ id X and, in this situation, Lemma 10.1 tells us that X is f -local, so (X, a) is L ′ -local. As proved in Lemma 9.1, the condition that T preserves f -equivalences and T f -equivalences in Theorem 10.2 is automatically satisfied for every f if the monad T is associated with a simplicial operad. Hence we infer the following general result. Corollary 10.3. If T is the reduced monad associated with a simplicial operad acting on pointed simplicial sets, then L f X ≃ L T f X for every map f if X is the underlying space of a T -algebra up to homotopy. Proof. This is a consequence of Theorem 10.2 and Lemma 9.1. Example 10.4. The infinite symmetric product SP ∞ is the reduced monad associated with the commutative operad. As shown in [28, Proposition 1.1], the homotopy algebras over SP ∞ coincide with the strict algebras -this is parallel to the fact that, in stable homotopy, the classes of homotopy HZ-module spectra and strict HZ-module spectra coincide, as proved in [44]. Corollary 10.3 implies that L f X ≃ L SP ∞ f X if X is a GEM, as pointed out in [28,Theorem 1.3] and generalizing [16, Corollary 3.2(iii)]. Hence, in particular, for every space X and every map f . Example 10.5. The homotopy algebras over the James functor J are the monoids in Ho(sSet * ) with its Cartesian monoidal structure, that is, the homotopy associative H-spaces. Since J is the reduced monad associated with the associative operad, we can infer from Corollary 10.3 that L f X ≃ L ΩΣf X for every homotopy associative H-space X and every map f between pointed connected simplicial sets. This is a more general statement than part (ii) of Corollary 9.7. It follows similarly that L f X ≃ L Ω ∞ Σ ∞ f X for every map f between pointed connected simplicial sets and every homotopy algebra X over the reduced monad Q associated with an E ∞ -operad. The homotopy Q-algebras are the H ∞ -spaces in the sense of [21, § I.3.7]. As shown in [56,64], H ∞ -spaces need not be homotopy equivalent to E ∞ -spaces. Thus we have also sharpened part (ii) of Corollary 9.8. More generally, for every cofibrant ring spectrum E, we consider the monad defined as T X = Ω ∞ (E ∧ Σ ∞ X) on Ho(sSet * ), and call its algebras unstable E-modules. for all unstable E-modules X. Proof. Let T X = Ω ∞ (E ∧ Σ ∞ X). Part (ii) of Proposition 5.1 applied to the adjunction Σ ∞ : Ho(sSet * ) ⇄ Ho(Sp) : Ω ∞ tells us precisely that Σ ∞ sends f -equivalences to Σ ∞ f -equivalences for every map f of pointed simplicial sets. Since E is assumed to be connective, smashing with E preserves Σ ∞ f -equivalences. This is proved using the derived function spectrum F(−, −) as in Section 8. Indeed, if E is connective, then so choosing Z to be Σ ∞ f -local yields our claim. Finally, Ω ∞ sends Σ ∞ f -equivalences to f -equivalences by part (i) of Corollary 9.8. Hence T preserves f -equivalences for every f , and Theorem 10.2 yields the desired result. Theorem 10.7. Let M be a model category and let T be a monad on M preserving weak equivalences and cofibrant objects. If A is a cofibrant object in M such that C A and C T A exist, and T preserves cofibrant A-cellular objects and cofibrant T A-cellular objects, then if X underlies a T -algebra up to homotopy. Proof. The proof follows the same steps as the proof of Theorem 10.2. Example 10.8. If A and E are cofibrant spectra and E is a homotopy ring spectrum, then, assuming either that E is connective or that C A commutes with suspension, the monad T X = E ∧ X preserves A-cellular objects and T A-cellular objects by the argument given in the proof of Theorem 8.1. It then follows from Theorem 10.7 that C A X ≃ C E∧A X if X underlies a homotopy left E-module. As a special case, C A X ≃ C HZ∧A X for every A when X is a stable GEM. Using Lemma 9.2, it follows from Theorem 10.7 that C A X ≃ C SP ∞ A X for every pointed connected simplicial set A if X is a GEM. Likewise, C A X ≃ C ΩΣA X if X is a homotopy associative H-space, and C A X ≃ C Ω ∞ Σ ∞ A X if X is an H ∞ -space. More generally, the analogue of Corollary 10.6 for cellularizations reads as follows. Corollary 10.9. Let E be a connective cofibrant ring spectrum. For every pointed connected simplicial set A we have C A X ≃ C Ω ∞ (E∧Σ ∞ A) X for all unstable E-modules X. Proof. The functor Σ ∞ sends A-cellular pointed simplicial sets to Σ ∞ A-cellular spectra by part (ii) of Proposition 6.3, while smashing with E preserves Σ ∞ A-cellular spectra since E is connective, and Ω ∞ sends those to A-cellular spaces by part (iii) of Corollary 9.9. Hence the monad T X = Ω ∞ (E ∧ Σ ∞ X) preserves A-cellular spaces and Theorem 10.7 applies.	2018-06-22T19:08:06.000Z	2014-04-29T00:00:00.000	{ "year": 2014, "sha1": "60fbaa047ec344bc0389f78629b7c1dd5f5372f4", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "1ec79af4d89b8490882881d636be916f518456b7", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics" ] }
17166271	pes2o/s2orc	v3-fos-license	How School Nurses Experience Their Work with Schoolchildren Who Have Mental Illness – A Qualitative Study in a Swedish Context Background: Reports from research have shown that mental illness has increased dramatically in recent years and is currently one of the biggest problems among Swedish children and adolescents. Aim: The aim of this study was to describe how Swedish school nurses experience their work with schoolchildren who have mental illness Method: Data were gained by individual interviews with school nurses (n = 10) and were analyzed by using manifest qualitative content analysis. Results: The results of the study showed that school nurses used various tools when working with schoolchildren who have mental illness. The working tools were regular health talks, motivational interviewing, individual counseling, family counseling, creating trust, and affirming the child’s confidence. Conclusion: Results of the study demonstrate the need for further research on schoolchildren’s experience of getting help and support from the school nurse. 2010b). The report View All of Me (Johansson & BRIS, 2012) showed that mental health problems in children often lead to consequences like deteriorating school performance, difficulty sleeping, feelings of loneliness, anxiety, self-hatred, shame, fear, guilt, self-destructive behavior, eating disorders, suicidal thoughts, difficulty controlling anger, and shying away from interacting with others (Alfvén & Alfvén, 2006;Cullberg, 2006). Student health has the mission to organize and provide health care for children and adolescents in school health services. All students in Sweden are entitled to student health, whether they attend a public or private school (National Board of Health and Welfare & Development Center for Children's Mental Health, 2010b); student health is offered and implied that they take up the offer. There is thus a risk that parents will reject this offer (National Board of Health and Welfare, 2004b). The new Education Act (Ministry of Education, 2010a), which came into force on 1 July 2011, replaces the concept of school health with student health and states in § 25 that there should be student health services for students in preschool classes, elementary school, compulsory school, Sami school, special school, secondary school, and upper secondary school. Student health includes medical, psychological, psychosocial, and socio-educational services. Student Health will be concerned primarily with prevention and health promotion. Pupils' progress toward the goals of education should be supported. For medical, psychological, and psychosocial interventions, students must have access to a school physician, school nurse, psychologist, and social worker (Ministry of Education, 2010a. The mental health among schoolchildren thus has declined dramatically in recent years, as discussed above, and is currently one of the biggest problems among Swedish children (Johansson & BRIS, 2012). Because only a few empirical studies based on Swedish school nurses' scientific perspective have touched on the subject, and because we need to gain a deeper understanding and greater knowledge of this problem, we considered it valuable to study the school nurse's experience of working with schoolchildren's mental health issues. The aim of this study was to describe how Swedish school nurses experience their work with schoolchildren who have mental illness. Design The study has used descriptive design with a qualitative approach (Polit & Beck, 2013). The first characteristic of qualitative researchers is that they often put together a complex collection of data obtained from a variety of sources and use different methods. This semi structured in depth research interview seeks to obtain uninterpreted descriptions (Kvale, 2007). The interviewed informant describes as accurately as possible what she is experiencing and feeling and how she acts. The key is to obtain comprehensive descriptions that reproduce the qualitative diversity and all the differences, variations, and complexities of an event (Kvale, 2007;Kvale & Torhell, 1997). For the present study are semi structured in depth interviews used. Context and Participants The survey was conducted in two medium-sized municipalities, located in southern Sweden, with a total of 62,413 inhabitants. The study group was nurses working in primary schools within the school health systems and includes students from preschool to grade six. The characteristics of these schools are that they are divided into primary and middle schools and are located in elementary schools. Another important feature is that all the schools are owned by municipalities. Although the distances were far between the various schools, in some schools schoolchildren of Swedish descent dominated, while in other schools children of foreign origin dominated. Inclusion criteria for participation in the study specified nurses who have specialist training as district nurses or in health care for children and young people and have worked at least one year as a school nurse. Both female and men who work as school nurses were included in the study. The author interviewed ten nurses. Of these, all were female. Their professional experience as nurses ranged from 1.5 years to 23 years, with an average time of 10 years spent as a school nurse. The school nurses were between 38 and 61 years old; the average age was 46 years. Eight nurses had specialized training in health care for children and adolescents and two nurses had specialized training as district nurses. The number of schools and students that school nurses were responsible for ranged from two to seven schools and to between 300 to 725 students. The purposive selection of informants in this study is appropriate (Polit & Beck, 2013). We consciously chose purposive sampling technique of people who would be most useful in the study. The selection was designed: (1) to find people who are appropriate to find some degree of interest in their work (2). An information letter explaining the purpose of the study was sent to two business managers for school nurses in two different municipalities in the southern part of Sweden to get written permission to conduct the study and to gain access to potential informants (nurses). One of two operations managers gave the author permission to conduct the study. The operations manager who did not give the author permission to conduct the study gave as a reason that school nurses' lack of time to participate in it. This led the author to broaden the application. Another operations manager in another municipality was asked for written authorization to conduct the study, and he gave the author permission. After the author received a written authorization from business managers to conduct the study, school nurses were contacted by telephone with and were informed about the purpose of the study and asked if they were interested in participating. In one municipality contacted by the telephone, seven of thirteen nurses met the inclusion criteria and agreed to participate. The loss was six nurses; two of them met the inclusion criteria but were unable to participate due to time constraints, one of them did not meet the inclusion criteria, and three of them did not respond at all, despite repeated phone calls and voice messages by phone. In the second municipality we contacted two nurses over the phone; both met the inclusion criteria and both agreed to participate, so the loss was zero. All nurses who agreed to participate were sent an information letter with written information and an invitation to participate in the study. The author booked the time and place of the interview in consultation with the school nurses who chose to participate in the study. Data Collection Method Data collection took place from October 2012 to January 2013. Data was collected through individual interviews based on an interview guide constructed with semi-structured questions and supplementary questions based on the study's purpose and questions (Polit & Beck, 2013). The items under investigation and the order in which they were addressed during the interview contained background questions about each school nurse's gender, age, education, work experience, and responsibility for the number of pupils/schools, among other questions. Specific questions asked were (1). Describe in as much detail and as much as possible how to work with schoolchildren who have mental illness. Supplementary questions included: What do you mean? Can you elaborate on that? Do you have anything more to add? A pilot interview was conducted with the result that the questions were tested on a school nurse so the author could ensure that the proper issues were addressed and that the answers responded to the purpose of the study. The pilot interview showed that no changes in the interview guide were needed. All interviews were conducted in the school nurse's office. The school nurses gave their approval by signing an informed consent form. The interviews were recorded on tape. The interviews were transcribed verbatim immediately after the interview so that nothing important was missed, which means that even pauses, laughs, or coughs were recorded. The length of the interviews varied between 15 and 25 minutes. Data Analysis Qualitative content analysis according was used to analyze the transcribed interviews. The focus in content analysis is to describe variations by identifying similarities and differences in text content. Differences and similarities are expressed in categories and themes at different levels of interpretation. In every text there is a manifest content and a latent message. The manifest content of the text closely reveals the content and is expressed on a descriptive level in terms of categories. Manifest analysis was chosen by the author to analyze the text of the transcribed interview area where the aim was to describe the school nurse's experience. The transcribed interviews were included in the analysis as an analysis unit (unit of analysis). The transcripts were read through several times by the first author (FD). First, a general, transparent reading was performed to get a sense of the whole; the next step was a reading to aim for a more thorough and in-depth understanding with the intent of getting ideas for further analysis with the purpose of the study as the basis. The text was then split into domains, namely, calls to identify and cooperate. A domain is part of the text that relates to a specific area and serves as a rough structure that is possible to identify with a low degree of interpretation. Meaningful units (meaning units) were picked out by the study's end. A meaningful unit may consist of words, sentences, and paragraphs of text that are related in content and context. During the analysis, the process was condensed and abstracted. Abstraction is a process that makes the text shorter and therefore more manageable, while the core content is preserved and nothing of significance disappears. Then the condensed text was abstracted, provided with codes, and subsequently merged into subcategories and categories. Ethical Considerations The study was performed in accordance with the Helsinki Declaration (World Medical Association, 2001) (Saif, 2000). All participants gave their informed consent to participate in the study after having been presented with detailed information about the study and their own participation. They were also informed that they had the right to terminate their participation at any time without consequences. Results The result of ten interviews with school nurses reported on the school nurse's work tools; the school nurse's experience identifying mental illness; and the school nurse's experience of cooperation with other professionals, agencies, and parents. The School Nurse's Working Tools The results of the study showed that the school managers used various work tools when dealing with schoolchildren's mental health. The school nurse used regular health talks, motivational interviewing, individual counseling, and family counseling. An important part of the school nurse's job was to create a trust and affirm the child's confidence. Health Talk The results of the study showed that health talks were an important tool for the school nurse. During the health talks school nurses use of "bear cards" that help the child to express different emotions. Family Counseling The study's results showed that working with the whole family and involving parents was usually important in dealing with mental illness in schoolchildren. The school nurses stressed that when it came to children from grade one to grade six, it was especially important to talk to the parents and to involve them when it was appropriate. "And then it's the parents--so it is about the smaller children now ... we cannot do anything without the parents, without parental consent ... you need them on the train the whole the time..." Creating Trust Many nurses felt that a quiet environment, security, and taking the time to listen led the children to find it easier to talk about their concerns and thus created trust in the school nurse. "... For my part, I think it feels like I need to give them peace of mind so that they might dare to open up a bit more about what it is that is their concern.". "…Be responsive if the child starts to talk, that you then take the time to listen… Most often it's the person who knows the child the best so they can open themselves..." To Affirm the Child's Self-Confidence Attempts to strengthen the child's self-confidence were demonstrated in the study. The school nurses could feel powerless at times because it was hard for them to make big changes. Then it was important to support and motivate the children to find their value in life and to equip them for the future. "These gray area kids ... You may try to run and motivate them to find their own value in life, and are they prepared for the future, that they should make their own wise choices." The school nurses found working in groups valuable for strengthening the child's self-confidence through values clarification. "Something that I think is good is working in groups in class, sometimes having different valuation exercises where you stand up for your opinions in front of others and ... We take up sensitive issues in class and in groups and how to do ... and I think they may be strengthened by such exercises." The School Nurse's Experience in Identifying Mental Health Issues Results of the study showed that all nurses had identified mental illnesses among schoolchildren such as agitation, anxiety, depression, and other neuropsychiatric disorders. Their experience in the profession and their feelings when a child is hurting them self physically helped them to be able to identify mental illness quickly in a child. The identification could be made in conjunction with a healthy conversation. Even when the child visited the school nurse several times and complained of little concerns such as sore feet, sore hands, and a headache or the child told me as well that for example he heard voices or made an explicit suicide threat. "... Many will hit themselves and tell. then when I make the right health calls, that's where I catch problems more efficiently, so to speak…" The parents could call the school nurse and tell the nurse they are worried about their children. Classroom teachers turned to the school nurse if they received signals that a child was not feeling well. Classmates, especially those in grades five and grade six, turned to the school nurse and told her they were worried about a friend they had noticed was not himself, was feeling sad, and more. Identification of mental health problems among schoolchildren led the school nurses to refer the child to the emergency room, the children's hospital, or to child and youth psychiatry for further investigation. "Anorexia ... where I know that the parents come to me and said, it does not look good; then I went and picked up the girl--no, it was not good. She had a weak pulse, her hands were cold, and I could not get any blood pressure; she was pale, tired, very rundown. Then I sent her to the children's emergency room, where they made a medical examination, and then she came to child and youth psychology." The School Nurse's most Difficult Experiences in the Identification of Mental Illness The results of the study showed that some nurses felt it was difficult when they were informed by the teachers that a child had a mental illness. The School Nurse's Experience of Cooperation with Other Professionals, Agencies, and Parents One of the study's results showed that one task with mental health problems among schoolchildren was the work with many different professions and agencies, such as school counselors, teachers and principals. Cooperation with child and youth psychiatry came when the school nurses sent a referral for a child who had a mental illness. Social Services came into play in connection with the notification that a child had been badly treated. Cooperation with Child House came when the nurse suspected that a child had been subjected to a violation by any adult or parent; the child was sent there and all the staff got there so that the police and the pediatrician could examine the child and an x-ray was done at the hospital. The collaboration worked fine; sometimes secrecy can cause problems, but the parents were with them so it worked mostly well. Cooperation with the parents was an important part of the work. Cooperation with School Counselor The school nurses work closest with the school counselor. The contact between the school nurse and school counselor was very close, despite different vocational skills, but they shared a certain sphere, they juggled their duties and skills with each other and greatly benefited from each other's expertise. Some schools had school counselors in place, at other schools counselors were available a few days a week, and at other schools there were no school counselors at all. In schools where counselors were not available, school nurses felt lonely and overworked; in situations where there was a counselor available, they did not have to juggle when a problem with a child arose. School nurses had more children visit with mental illness. "... The cooperation that I have such has been with the school counselor it is the most often between us, it's nothing that we're talking about with the teams ... but I almost think that the best collaboration is that it works directly between the school nurse and the school counselor, who do not go through the team ". Cooperation with Student Health Team All nurses experienced working with the student health team. The student health team included the school nurse, school counselor, principal, the special education team, social educator, psychologist, and sometimes even teachers and speech therapists. The student health team met and discussed children who feel poor. The discussion about a child who was feeling poorly concerned how to proceed, how to help the child, how to help the family, whether help from child psychiatry and social services was needed. It was helpful when the student health team met frequently, at least a couple times a month. "... It is interesting sometimes when you are sitting in the student health team with the principal, counselor, special education teacher, psychologist and speech therapist at times and that they have different skills that they can use to look at the case from different angles... it's rewarding that it goes beyond school nurses and pediatric nurses and offers a multifaceted view of the students..." Cooperation with Parents and Other Parties through Meetings Among the study's results, it appears that school nurses had experience working with parents and other agencies also through meetings. People from school, child and youth psychiatry, social services, children's rehabilitation, Discussion of Results Results of the study showed that school nurses used various tools when working with schoolchildren who have mental illness. It was revealed that all nurses had been through and identified the mental health of schoolchildren. It also emerged that in the work of mental health problems among schoolchildren it is important to work with many different professionals, agencies, and parents (Barnard & Neal, 1977;Browne, Cashin, & Graham, 2012) (Kim, 2010). According to the National Board (2004b), health talks by school nurses developed a special form of support and a health-promotion approach, based on individual strengths and weaknesses, which is in line with the results of the study. The results of the study showed that the school used various tools when working with schoolchildren's mental health in which health talks were an important tool (Browne et al., 2012). During the health talks, school nurse used bear cards, for example, that helped the child express various emotions. Another study (Golsäter, Sidenvall, Lingfors, & Enskär, 2011) described the schoolchildren's health as an opportunity to get the support of the school nurse. Healthy speech was perceived as an opportunity to establish contact with the school nurse and try to find out if she was someone to be trusted or not (Lundquist, 2008). Working with the whole family and involving the parents was often important in mental illness in schoolchildren, the study's results showed. This is supported by several studies E. Clausson et al., 2003;E. K. Clausson et al., 2008aE. K. Clausson et al., , 2008b. The school nurse listened to the child and then contacted the parents and told them about the situation and asked them to come to her office to discuss the situation (Pidgeon, 1985;Pryjmachuk, Graham, Haddad, & Tylee, 2012). By meeting together with the child, the parents, and the school nurse, the school nurse could help the child to express the problem in words (Morberg, Lagerstrom, & Dellve, 2012). Another study highlighted the importance of therapeutic conversations with families who had children who suffered from mental illness (Thompson, 1977). Therapeutic conversations seemed to trigger a healing process, including affective, cognitive, and behavioral changes in family function where the families became more and more aware of their own resources and capabilities to manage their health problem (Maenpaa & Astedt-Kurki, 2008). The meetings led to feelings of relief; families felt affirmed and created their own solutions . In was shown that the most common way for the nurse to pay attention to mental health problems among the schoolchildren was through health talks, which is in line with the results of the study that stated that the identification of mental health problems among schoolchildren could be accompanied with a healthy conversation (National Board of Health and Welfare & Development Center for Children's Mental Health, 2009). A study by Clausson, Kohler, and Berg (2008) showed, however, that the health checks and health talks were assessed and declared a common basis for assessing the physical health of the schoolchildren and less commonly to assess mental health. Among the study's results was the revelation that school nurses could identify mental illness in schoolchildren even when the child visited the school nurse several times and complained of little concerns like sore feet, sore hands, and a headache. Schoolchildren's spontaneous visits to the school nurse seemed to be more common and were warranted to assess schoolchildren's mental health. Study results also showed that the identification of mental health problems among schoolchildren led to school nurses referring the child to the children's emergency room, a children's clinic, or child psychiatry for further investigation. This is supported by a study by Moberg and colleagues (2012), which identified school nurses as "gatekeepers," identifying first-line problems and contacting parents and other professionals if necessary. Among the study's results was the demonstration the work with mental health problems among schoolchildren was part of school nurse's duties, including collaboration with many different professionals, agencies, and parents. It was revealed that school nurses had experience in working with parents and other parties through meetings that were very important because they all were working toward the same goal, namely, to help the child. Moberg and colleagues (2012) described in their study of school nurses' consideration of themselves as advocates for the schoolchildren; when a child appeared to be suffering from mental illness, she could assess the www.ccsenet.org/gjhs Global Journal of Health Science Vol. 6, No. 4;2014 child's school situation and the school environment, talk to counselors, psychologists, teachers, and most importantly, involve the child's parents. In a study by Pryjmachuk and colleagues (2012), parents were seen as important and needy when a schoolchild showed mental illness, and often they wanted immediate change in their child's behavior. Another study showed that parents were willing to participate in their children's health examinations and they wanted nurses to invite them often. When the parents were invited, it offered them an opportunity to express their own views about the child's health. Parents experienced such meetings as relaxed and friendly (Maenpaa & Astedt-Kurki, 2008). Conclusion In summary, the school nurse's experiences of identifying mental health problems among schoolchildren as well as their experiences working with other professionals, agencies, and parents refers to the nurse's cognitive, behavioral, and social aspects of professional action. The results of the present study show that the school nurse have an important role in reasoning, making decisions, transferring knowledge into action, putting available knowledge into practice and taking certain actions. In conclusion, this study shows that the school nurse has a central role in the work of schoolchildren's mental health. Results of the study raise concerns about further research on schoolchildren's experiences getting help and support from the school nurse.	2016-10-08T01:47:31.943Z	2014-03-17T00:00:00.000	{ "year": 2014, "sha1": "5e2f893d1535cdb4a8e7fb85f284f0cd93ed055c", "oa_license": "CCBY", "oa_url": "http://www.ccsenet.org/journal/index.php/gjhs/article/download/34110/19966", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "5e2f893d1535cdb4a8e7fb85f284f0cd93ed055c", "s2fieldsofstudy": [ "Education", "Medicine", "Psychology" ], "extfieldsofstudy": [ "Medicine" ] }
198255414	pes2o/s2orc	v3-fos-license	Association of Sperm Aneuploidy Frequency and DNA Fragmentation Index in Infertile Men. Background For improving the evaluation of male infertility, many parameters were studied and reported in earlier literature. The aim of this study was to estimate the frequency of sperm aneuploidy and DNA fragmentation in infertile men and to assess the correlation between sperm aneuploidy and DNA fragmentation. Methods In this study 100 infertile men were included, cases with azoospermia were 68%, oligospermia 18%, severe oligospermia 6%, and oligoasthenoteratospermia (OAT) 8%. Ten normozoospermic men who had two normal children were included as a control. The sperm aneuploidy test by Fluorescence In Situ Hybridization (FISH) and sperm DNA fragmentation index by TdT (Terminal deoxynucleotidyl transferase)-mediated dUTP nick end labelling (TUNEL) were carried out. To determine the aneuploidy status and DNA fragmentation index, frequency was used. The correlation between sperm aneuploidy and sperm DNA fragmentation along with age was assessed by using Spearman's correlation coefficient. The p<0.05 was considered significant. Results The age of 100 subjects ranged between 22-48 years (35.5±5.1). Sperm aneuploidy frequency and DNA fragmentation rate were found to be higher in infertile men compared to control men (n=10). There was a significant relationship between age and sex chromosomal aneuploidy (p<0.05) and significant difference between sperm aneuploidy and damaged DNA (p<0.05). Conclusion FISH and TUNEL assay results showed increased sperm aneuploidy frequency, and DNA fragmentation index in infertile men compared with the fertile men. There is significant relationship observed between sperm aneuploidy and DNA fragmentation. These two parameters are important and they must be investigated for clinical practice. Introduction nfertility is a relatively significant problem that affects approximately 1 in 6 couples worldwide (1). Among them, male factor contributes approximately to 50% of the cases (2,3). Wide ranges of factors are involved in male infertility, among which genetic factors play a major role in few cases. The regular semen analysis measures only the sperm production and sperm quality which can not reveal the reason for infertility even with normal semen parameters. A single definite factor influencing male infertility is the sperm DNA integrity. The possible sperm nuclear variations are abnormal chromatin structure, Y chromosome microdeletion, sperm aneuploidy and DNA fragmentations (3). Non-disjunction and anaphase lag are the two mechanisms causing chromosome segregation in meiosis, primarily the non-disjunction leading to production of aneuploid sperm during spermatogenesis in human (4). The common treatment for severe male factor in-JRI fertility was intracytoplasmic sperm injection (ICSI) and increased incidence of sex chromosomal aneuploidy in offspring was evident by this method due to bypassing the natural selection criteria (5). During the process of chromatin remodeling, unrepaired DNA breaks were generated as a result of defective spermiogenesis, and oxidative stress could be the other mechanism for DNA damage (6). Sperm aneuploidy and sperm DNA fragmentation in infertile men is more frequent compared to the general population, which needs to be understood well (6,7). Therefore, our main aim was to evaluate the rate of sperm aneuploidy and DNA fragmentation in infertile men to justify the inclusion of these parameters in routine practice. Study population: This study was performed on 100 infertile men with abnormal semen parameters who were referred to KSHEMA Centre for Genetic Services, K.S. Hegde Medical Academy from 2014 to 2018. Among them, cases with azoospermia were 68%, oligospermia 18%, severe oligospermia 6%, and oligoasthenoteratospermia (OAT) 8%. Ethics committee approval was obtained from the Institutional Ethics committee. Sperm aneuploidy test by Fluorescence In Situ Hybridization (FISH) and sperm DNA fragmentation test by TdT (Terminal deoxynucleotidyl transferase)-mediated dUTP nick end labelling (TUNEL) were carried out on semen samples of patients with oligospermia, severe oligospermia and OAT patients (32%). Among 32 infertile men, 14 (Severe oligospermia and oligospermia (64.7%) and oligoasthenoteratospermia (35.3%)) were willing to give their semen samples. Informed consent was obtained from all the participants. Normozoospermic men who had two normal children were considered and included as a control for sperm aneuploidy and sperm DNA fragmentation index. Processing and pretreatment of semen samples for sperm aneuploidy: Semen samples were collected by masturbation in a sterile container after 3-5 days of sexual abstinence and left at room temperature for 30 min. Processing of semen sample was carried out as previously described by Bernardini et al.'s (8) protocol with modifications. 20 ml of Phosphate Buffer Saline (PBS) was added and centrifuged at 2000 rpm for 10 min and repeated for two more times to get the sperm alone. The supernatant was discarded, and 20 ml of pre-warmed hypotonic solution (0.075 M KCl) was added and incubated for 13 min at 37C. After incubation, 1 ml of pre-cold Carnoy's fixative was added and centrifuged at 2000 rpm for 10 min. Final fixation was done with 10 ml of fixative by using vortex and kept in the freezer for 30 min. The fixative change was given until obtaining a clear white pellet. Sperm aneuploidy by FISH protocol: FISH was standardized based on Sarrate et al. (9) with slight modifications using FAST FISH prenatal enumeration probe kit (Cytocell, UK) to detect the aneuploid cells frequency of spermatozoa in infertile men. This kit includes the centromeric probes to study chromosomes X, Y, 18 (Green, Orange, and Blue, respectively) and unique sequence for chromosomes 13 and 21 (Green and Orange). Slide preparation: Two drops of concentrated cell pellet were dropped onto a marked area of the slide and air dried. The slides were left at room temperature for 1 hr. Pretreatment of the slide: The prepared slides were treated with 2x sodium saline citrate (SSC) for 5 minutes. Then slides were transferred into coupling jar containing DTT (25 mM) (Dithiothreitol) and treated for 10 min. The slides were placed in PBS for 2 min at room temperature and dehydrated with ethanol gradient of 70%, 85% and 100% for 2 min each and completely dried at room temperature. Then, the slides were treated with SSC (2x) again for half an hour and dehydrated with ethanol gradient of 70%, 85% and 100% for 2 min each. Co-denaturation: Slides were dried at room temperature, and 4 µl of FAST FISH prenatal enumeration probe kit (Cytocell, UK) was added. They were covered with cover-slip and the coverslip was sealed with rubber cement. The slides were then placed in a Start Spin ThermoBrite plate overnight. Denaturation at 75C for 5 min and hybridization at 37C for 18 hr were done by the automated program in the ThermoBrite machine. Post-hybridization washes: A coupling jar containing SSC (0.4x)/0.05% Triton x100 was placed in a 73±1C water bath, 30 min before use. Rubber cement and coverslips were removed carefully and the slide with SSC (0.4x)/Triton x100 was treated for 1 min. Then, the slides were treated with SSC (2x)/Triton x100 for 5 s at room temperature. The slides were air dried and 5 µl of Arumugam M, et al. JRI DAPI (4,6-diamidino 2-phenylindole) was applied in the target area and immediately covered with a coverslip and sealed. Enumeration and analysis of slides: The slides were observed using a suitable filter set on a BX 53 Olympus fluorescence microscope, and the signals present in each sperm were enumerated and captured using a CCD camera attached with the microscope. Good images were captured and documented using the GENASIS Version 7.2 Software. Sperm aneuploidy and diploidy rates for chromosome 13, 21, 18, X and Y were observed and recorded. In sperm, FISH signals were scored based on previously described criteria (2). In brief, the single signal in individual sperm represents a normal signal pattern for an exact number of chromosomes ( Figures 1A and B). If additional signals were present in the cell, they were disomic for autosomes ( Figure 2A) and diploid for sex chromosomes ( Figure 2B). When the sperm contains two signals for each chromosome, then it is considered as diploidy. When there are no signals on sperm, it is considered as nullisomic, and nullisomic sperm could be observed in all the samples analyzed. DNA fragmentation index: Terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate (dUTP) In Situ DNA nick end labelling (TUNEL) assay was performed with a slight modification of sperm suspension after density gradient separation as previously described (10). A part of semen sample from the control group and patients were washed with 20 ml of PBS and spun at 2000 rpm for 10 min. The supernatant was discarded, and the cells were resuspended with 4% paraformaldehyde fixation buffer and permeabilized with 0.25% Triton X-100 in PBS. The supernatant was discarded and 1 ml of cold 70% (v/v) ethanol was added and stored at -20C until further process. DNA strand breaks were detected by using a commercially available kit (Click-iT Plus TUNEL assay for In Situ apoptosis detection with Alexa Fluor dyes, molecular probes, life technologies) according to the manufacturer's instructions. Positive control was performed with 1 unit of DNase I diluted into DNase I Reaction Buffer (x 1) for 30 min at room temperature. After incubation, the slides were washed with deionized water and further proceeded to TdT reaction. The percentage of spermatozoa with fragmented DNA was determined by direct observation of 500 spermatozoa per sample with Olympus BX 53 fluorescence microscope. The sperm cells were counterstained with Hoechst 33342 (Blue). Fluor 488 picolyl azide dye (Green) was used to detect the sperm with TUNEL positive strand breaks. Sperm with DNA fragmentation were clearly visible in green color (Alexa), and sperm without DNA fragmentation were in blue color (Hoechst 33342) (Figure 3). Statistical analysis: The collected data were summarized by using frequency/percentage for qualitative data and mean with standard deviation for quantitative data. To determine the aneuploidy Association of Sperm Aneuploidy and DNA Fragmentation JRI status and DNA fragmentation index, frequency was used. To compare the mean of sperm aneuploidy and DNA fragmentation rate, student t-test was used. The correlation between sperm aneuploidy and sperm DNA fragmentation along with age was assessed by using Spearman's correlation coefficient. The p<0.05 was considered statistically significant. Data management and analysis were performed by using Microsoft Excel and Statistical Package for the Social Sciences v16.0.1 (SPSS Inc, Chicago, IL, USA). Results The age of 100 subjects ranged between 22-48 years with a mean and standard deviation as 35.5± 5.1. Correlation between sperm aneuploidy and DNA fragmentation with age is given in table 1. A significant correlation between age and sex chromosomal aneuploidy was observed (r=0.363, p<0.05). Sperm aneuploidy frequency (10.1±10.8) was found to be higher in infertile men as compared to fertile men. Increased disomy frequency of chromosome 13 and XY disomy was observed. The DNA fragmentation rate (61.6±21.9) was comparatively high in infertile men than fertile men. The results in table 2 indicate a positive significant difference between sperm aneuploidy and sperm with fragmented (Damaged) DNA (p<0.05); there is no significant difference between sperm aneuploidy and sperm without damaged DNA. Discussion Our observation showed the percentage of sperm aneuploidy increases by the age of infertile men, mainly on sex chromosomal aneuploidy. Wyrobek et al. (11) had the largest investigation and observed the major association between age and frequencies of aneuploidy and diploid sperm. A study conducted by Martin et al. (12) observed no correlation between paternal age and "sex ratio" in sperm and aneuploidy frequency. The reason for the increased miscarriages and abnormal fetuses is abnormal sperm aneuploidy and the most widely used protocol to estimate the sperm aneuploidy frequency is FISH (13). Our results are similar to those reported by Aran et al. (14) who found an increase in sex chromosome disomy and diploid spermatozoa in infertile men. Moosani et al. (15) observed a significant increase in the frequency of disomy for chromosome 1 and XY disomy by FISH and they also observed increased frequencies of numerical abnormalities by sperm karyotyping. In our study, increased disomy frequency of chromosome 13 and XY was observed. Our results showed an increased rate of DNA fragmentation in infertile men compared to control men, not with age. A study conducted by Singh et al. (16) revealed an increased effect of DNA damage with age and decreased age-related apoptosis in human sperm. Sperm DNA fragmentation by TUNEL assay is a good predictive parameter by its sensitivity and specificity, but not an independent measure of sperm quality (17). The main outcome of this study indicates a positive significant correlation between sperm aneuploidy and sperm DNA fragmentation. A study conducted by Di Santo et al. (2) manifested that among 109 infertile men, a significant positive correlation existed between sperm DNA fragmentation and sperm aneuploidy and no significant correlation was found with normal men. In our study, similar results were observed. There are previous studies by different authors which confirm the same result (5)(6)(7)18). There was a positive significant correlation between the sperm chromosome abnormality rate by FISH and DNA fragmentation even by using a combined method of Sperm Chromatin Dispersion (SCD) test (19). Another study on four infertile men who were carriers of balanced chromosomal abnormalities has shown the similar results (7). However, no significant correlation between DNA fragmentation and sperm aneuploidy was found by Balasuriya et al. (20). This type of inconsistent results explained that different tests were used to evaluate the sperm DNA fragmentation; there was a different number of probes to detect sperm aneuploidy Conclusion In conclusion, it can be concluded that the sperm aneuploidy and sperm DNA fragmentation are important parameters and they may be suitable for clinical practice. The reason that sperm assay is not carried out in routine practice is due to the lack of standardized protocols. By the current study, these techniques were standardized and could be used for regular diagnosis. High frequency of sperm aneuploidy and sperm DNA fragmentation might contribute to low fertilization rate and poor pregnancy outcome. There is significant relationship observed between sperm aneuploidy and DNA fragmentation. These two parameters are important and has to be investigated separately for clinical practice. However, large scale studies with specific infertile men as the subgroup may have benefit for specific therapeutic management.	2019-07-26T08:08:03.123Z	2019-06-25T00:00:00.000	{ "year": 2019, "sha1": "d33a13f2b41296452ca5bb48c9356724bca8f76d", "oa_license": "CCBY", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "ca8cd438b9ec6acd582596ced48270879efa0cd8", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
270919535	pes2o/s2orc	v3-fos-license	SPARC overexpression in allogeneic adipose-derived mesenchymal stem cells in dog dry eye model induced by benzalkonium chloride Background Nowadays, companion and working dogs hold significant social and economic importance. Dry eye, also known as dry keratoconjunctivitis (KCS), a common disease in ophthalmology, can readily impact a dog’s working capacity and lead to economic losses. Although there are several medications available for this disease, all of them only improve the symptoms on the surface of the eye, and they are irritating and not easy to use for long periods of time. Adipose-derived mesenchymal stem cells (ADMSC) are promising candidates for tissue regeneration and disease treatment. However, long-term in vitro passaging leads to stemness loss of ADMSC. Here, we aimed to use ADMSC overexpressing Secreted Protein Acidic and Rich in Cysteine (SPARC) to treat 0.25% benzalkonium chloride-treated dogs with dry eye to verify its efficacy. For in vitro validation, we induced corneal epithelial cell (HCECs) damage using 1 µg/mL benzalkonium chloride. Methods Fifteen male crossbred dogs were randomly divided into five groups: normal, dry eye self-healing control, cyclosporine-treated, ADMSC-CMV-treated and ADMSC-OESPARC-treated. HCECs were divided into four groups: normal control group, untreated model group, ADMSC-CMV supernatant culture group and ADMSC-OESRARC supernatant culture group. Results SPARC-modified ADMSC had the most significant effect on canine ocular surface inflammation, corneal injury, and tear recovery, and the addition of ADMSC-OESPARC cell supernatant also had a salvage effect on HCECs cellular damage, such as cell viability and cell proliferation ability. Moreover, analysis of the co-transcriptome sequencing data showed that SPARC could promote corneal epithelial cell repair by enhancing the in vitro viability, migration and proliferation and immunosuppression of ADMSC. Conclusion The in vitro cell test and in vivo model totally suggest that the combination of SPARC and ADMSC has a promising future in novel dry eye therapy. Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1186/s13287-024-03815-z. Introduction In recent years, the demand for eye care has been rising in China due to rapid economic development and significant lifestyle changes.This has resulted in a substantial increase in the incidence of ocular diseases.Concurrently, veterinary clinics have been witnessing a high prevalence of ophthalmic diseases among animals [1].The social and economic development, coupled with urbanization, has led to an improvement in people's spiritual, cultural, and material lives.Additionally, the emergence of objective environmental factors, such as aging families, has made pet dogs an essential source of emotional support for individuals [2].Moreover, dogs serve various functions, including search and rescue, drug detection, guiding, and companionship, making their economic value immeasurable [3].Consequently, safeguarding the health status of dogs is crucial to ensure their working effectiveness and to realize social and economic benefits.One of the most prevalent canine ophthalmic diseases is dry eye syndrome, with its incidence rate steadily increasing over the years, as revealed by research findings. Dry eye, also known as Keratoconjunctivitis Sicca (KCS), is a disease characterized by inflammation, pain, redness, and itching in the eye [4].It is caused by abnormal tear quality or quantity, as well as damage to the eye's surface resulting from unstable lacrimal secretion.The main symptoms of dry eye include dryness, a foreign body sensation, and a burning sensation in the eye [5].In the early stages, the eye's surface becomes cloudy and dry before progressing to fibrosis and even blepharospasm. Another common cause of dry eye is Sjögren syndrome, an unexplained chronic systemic autoimmune disease that affects the exocrine glands, resulting in dryness of the mouth, eyes, and other mucous membranes due to lymphocytic infiltration and glandular dysfunction [6].Therefore, the pathogenesis of dry eye is complex, with multiple etiologies.Current research suggests that the localized inflammatory immune response of the eye plays a significant role in the pathological damage associated with dry eye.Consequently, anti-inflammatory treatment is a prominent area of investigation [7].Although various drugs, such as the immunosuppressant cyclosporine, are available for KCS, they are irritating and not suitable for long-term use [8].Conversely, adipose-derived mesenchymal stem cells (ADMSC) have the advantage of weak immunogenicity and do not actively release their identity information once they enter the host.By participating in the immune process, they can evade the host's immune response through immune privilege.Moreover, ADMSC can exert regulatory effects on various aspects of the immune response, which gives them an edge in treating dry eye compared to immunosuppressants [9].However, there are several critical issues associated with ADMSC as a treatment option.The cellular state and activity of ADMSC are influenced by unfavorable factors in the in vivo environment and the outside world, resulting in reduced therapeutic effects.Prolonged in vitro culture leads to decreased proliferative capacity, senescence, and morphological changes of ADMSC.After transplantation, approximately 90% of the ADMSC die within 72 h, and their survival and proliferation rates are not as high as desired.This can be attributed to the lack of required nutrients or growth factors in the body.Although some studies have attempted to enhance the viability of ADMSC in both in vivo and in vitro settings, the results have been unsatisfactory [10]. SPARC (Secreted Protein Acidic and Rich in Cysteine), also known as Secreted Protein Acidic Collagen Osteoproteins, was first identified by Termine et al. in bone tissue in 1981 [11].This protein, rich in cysteine and containing osteocalcin and BM-40, is a 43 kDa protein secreted into the extracellular matrix by various cell types [12].It plays a significant role in regulating the extracellular matrix by participating in various processes, including cell adhesion, cell migration, cellular differentiation, and matrix remodeling [13].Several key findings regarding the functional role of SPARC have been elucidated in the literature.Nie J et al. demonstrated that SPARC modulates WAT body composition by influencing ASC mobilization through the α5β1 integrin complex [14].Moreover, Bradshaw AD et al. highlighted the impact of SPARC expression on the cell structure of mesangial cells, in addition to its involvement in regulating the cell cycle of fibroblasts and smooth muscle cells [15].Importantly, SPARC has been shown to be particularly abundant during cell construction of peripheral structures and tissue remodeling, notably during the remodeling and repair of adult tissues such as chondrocytes and osteoblasts [16].Furthermore, studies indicate that SPARC plays a regulatory role in the activity of key growth factors including platelet-derived growth factor (PDGF), fibroblast growth factor (FGF), and vascular endothelial growth factor (VEGF).These growth factors are intricately involved in diverse biological processes such as embryonic development, angiogenesis, tissue remodeling, and cell renewal [17].Moreover, it has been observed that the expression of SPARC is regulated during corneal injury and repair.It is involved in the migration and regeneration of corneal epithelial cells, as well as a series of inflammatory responses.Jessica Feldt et al. showed that SPARC, a novel blastocyte glycoprotein, is expressed by lacrimal myoepithelial cells.The lack of SPARC in adulthood may impair myoepithelial cell contraction and tear secretion [18].Consequently, its involvement is closely associated with the occurrence and development of ocular diseases such as dry eye. Thus, the potential of using SPARC gene-modified ADMSC to treat dry eye is a significant scientific inquiry that merits investigation.This study examines the impact of ADMSC with SPARC overexpression on the treatment of benzalkonium chloride-induced dry eye in canines [19,20]. Animals All animal experiments and conducted procedures were in accordance with the law on animal experimentation and are approved by the regulatory authorities.The work has been reported in line with the ARRIVE guidelines 2.0. To establish the dry eye-dry keratoconjunctivitis model, twelve 1-year-old healthy male crossbreeds weighing 2.0 ± 0.5 kg were obtained from the Experimental Animal Center of Northwest Agriculture and Forestry University.Additionally, three 1-year-old healthy male crossbreeds weighing 2.0 ± 0.5 kg were utilized as the normal control.For the ADMSC isolation, a 6-month-old female hybrid dog weighing 1.5 kg was obtained from the Laboratory Animal Center of Northwest A&F University.All animal experimental protocols were conducted in strict accordance with the Guide for the Care and Use of Laboratory Animals (Ministry of Science and Technology of the People's Republic of China, Policy No. GB/T35892-2018).The animals were housed in routine sanitary facilities with the required constant temperature and relative humidity. Cell isolation and culture ADMSC were derived from the abdominal subcutaneous adipose tissue of a 6-month-old female crossbred dog, as female dogs generally have more adipose content.The detailed ADMSC isolation steps and the identification of ADMSC and the establishment of the ADMSC immortalized cell line were described in our previous report [21].Cells were cultured at 37 °C and 5% CO 2 incubator in α-MEM (Invitrogen, Carlsbad, CA, USA) complete medium with 10% FBS (Gibco original, origin Australia).When cells were attached to the bottom of the plate at approximately 80% density, a 1:3 passaging was performed.Cells were frozen using DMSO cell freezing solution (Beyotime) [22,23].We used cells that were passaged to the third generation after resuscitation in all subsequent experiments. Establishment of ADMSC cell lines overexpressing SPARC The lentiviral expression vector for SPARC was reconstructed by inserting SPARC into the polyclonal site (CMV) of the pCDH-CMV-MSC-EF1 vector.Subsequently, this plasmid was transfected into 293T cells in the presence of two helper plasmids (PAX, VSVG) for lentiviral packaging.After 48 h of viral tapping, the cell supernatant was collected and the virus particles were purified and concentrated.The concentrated virus particles were then tapped into immortalized ADMSC cell lines for further experimentation.Green fluorescence-positive cell clones were screened using 1 µg/ mL Puromycin, and single-cell clones displaying green fluorescence were selected through the dilution method.Finally, the selected clones were expanded in culture to measure mRNA levels using real-time quantitative PCR and protein levels using ELISA [24]. Real-time fluorescence quantitative polymerase chain reaction analysis Total RNA was extracted from ADMSC-CMV and ADMSC-OESPARC using TRIzol reagent (Takara, Japan) according to the instructions provided by the reagent vendors.The extracted RNA was then reverse transcribed to cDNA using a reverse transcriptase kit (Thermo Fisher Scientific).Subsequently, a quantitative real-time polymerase chain reaction (qRT-PCR) was carried out using a CFX96 real-time PCR system.The qRT-PCR protocol involved pre-denaturation at 94 °C for 5 min, denaturation at 94 °C for 30 s, annealing at 58 °C for 30 s, extension at 70 °C for 30 s, and a total of 39 cycles.As an internal reference, GAPDH was used.The relative expression of genes was evaluated through the utilization of comparative CT values obtained from the qRT-PCR.The following primer sequences were used: GAPDH, F: G [25]. Detection of SPARC protein levels in cells by ELISA SPARC levels in ADMSC-CMV and ADMSC-OESPARC supernatants were determined by double-antibody sandwich assay using the enzyme-linked immunoassay 96T kit for cysteine-rich acidic protein (SPARC), and five replicate wells were set up for each group.The assay was performed according to the instructions provided by the reagent supplier (FANKEWEI, Shanghai, China).Absorbance (OD) was measured at 450 nm using an enzyme counter and the concentration of SPARC in the samples was calculated from the standard curve [26]. Cell growth curves Both ADMSC-CMV and ADMSC-OESPARC groups of cells were inoculated into 24-well plates at 5 × 10 3 cells per well.The medium in the plates was changed daily using α-MEM (+).Cells were manually counted in 3 wells of each group every 24 h.This process was repeated until day 8. Finally, based on the cell count results, a cell growth curve was plotted, with the horizontal coordinate indicating time and the vertical coordinate indicating the number of cells [27]. EdU cell proliferation assay Cells in logarithmic growth phase were used in this study.First, ADMSC-CMV and ADMSC-OESPARC cells were inoculated into 96-well plates with approximately 1 × 10 3 cells per well.The plates were then cultured until reaching a density of 60-70%.After discarding the culture medium, EdU staining was performed based on the instructions provided by the reagent vendor (Reebok Bio, Guangzhou, China).The staining process was observed and photographed under a fluorescence microscope immediately upon completion.Multiple fields of view were randomly selected for each sample, followed by the counting of EdU fluorescence-positive cells and calculation of the percentage of positive cells using Image J [28]. Models of canine dry eye disease After a 3-week acclimatization period, fifteen male crossbred dogs were randomly divided into five groups: normal, dry eye self-healing control, cyclosporine-treated, ADMSC-CMV-treated, and ADMSC-OESPARC-treated, three dogs in each group.To model canine dry eye, all dogs, except the normal group, were given 0.25% benzalkonium chloride drops in both eyes twice a day for seven consecutive days [29].From day 8 onwards, different treatments were administered to each group.The dry eye self-healing control dogs received 0.7% saline drops in both eyes once a day for eight consecutive days.The cyclosporine treatment group received cyclosporine ophthalmic solution in both eyes once a day for eight consecutive days.The ADMSC-CMV treatment group received 200 µL of 0.7% saline suspension containing 1 × 10 5 ADMSC in both eyes once a day for eight consecutive days.Similarly, the ADMSC-OESPARC treatment group received 200 µL of 0.7% saline suspension containing 1 × 105 ADMSC in both eyes once a day for eight consecutive days [30,31].Each time the 0.7% saline cell suspension was freshly prepared before using, with no more than a 15-minute interval between isolation of the cells from the petri dish and their use on the surface of the experimental dog's eye. Ocular surface inflammation index Ocular surface inflammation was observed in dogs at eight time points: day 1, day 3, day 5, day 7, day 9, day 11, day 13, and day 15 of the experiment.The dogs were anesthetized with 1% sodium pentobarbital 0.1 mL/ 10 g intraperitoneally and kept sedentary to facilitate the detection of various dry eye indices.Ciliary congestion (absent, 0; present but less than 1 mm, 1; present and greater than 1 mm less than 2 mm, 2; present and greater than 2 mm, 3); central corneal edema, peripheral corneal edema and peripheral corneal edema (absent, 0; present but clear iris texture can be seen, 1; present but clear iris texture cannot be seen, 2; present but pupil cannot be seen, 3).The sum of the three scores is the final Keratoconus Inflammation Index, and it is important that the same person who don't know the grouping situation performed and scored each operation [32,33]. Corneal sodium fluorescein staining The corneal fluorescein sodium staining method is commonly used in clinical practice to evaluate the integrity of the corneal epithelium.A positive result indicates a corneal epithelial defect, suggesting discontinuity of the corneal epithelial cell layer.Corneal sodium fluorescein staining experiments were performed at five time points: day 1, day 3, day 7, day 9, and day 15 of the experiment.To perform the staining, 0.5 mL of 2% liquid fluorescein sodium was pipetted onto the surface of the canine eye using a 1 mL spiking gun.After 2 min, corneal epithelial staining was observed under cobalt light using a slit lamp, and photographs were taken to record the staining.The cornea was divided into four regions: supratemporal, infratemporal, supranasal, and infranasal.A scoring system ranging from 0 to 4 was used to evaluate each region's staining.The scoring criteria were as follows: no staining (0 points), scattered punctate staining (1 point), slight diffuse punctate staining (2 points), severe diffuse staining but not in obvious lamellae (3 points), and obvious lamellar staining (4 points).After scoring was completed for each region, the scores were totaled.It is important that the same person who don't know the grouping situation performed and scored each operation [34]. Tear secretion experiments Tear volume was measured at four time points: day 1, day 7, day 10, and day 15 of the experiment.In order to ensure accurate measurements, precautions were taken to avoid the use of eye drops and bright light stimulation in the examination room prior to the assessment.The measurement technique involved using a 40 mm×5 mm strip of Whatman 41# filter paper.This strip was carefully folded at a right angle using sterile forceps and then clamped inside the conjunctival sac at the inner 1/3 of the lower eyelid.The other end of the strip was hung on the outside of the lower eyelid and left in place for 2 min.After this time, the wet length of the strip was observed and recorded.According to the established criteria, a wet length of 15 mm to 25 mm was considered normal, while a wet length of 11 mm to 14 mm indicated early dry eye.A wet length of 6 mm to 10 mm was classified as moderate dry eye, and a wet length of ≤ 5 mm was indicative of severe dry eye [35]. Collection of canine serum On the 15th day of experiment, serum was collected from all groups of dogs.To collect the serum, blood was drawn from the cephalic vein of the forearm.Before drawing the blood, the hair in the area was clipped, and the skin was sterilized with iodine and alcohol.The blood collector then tightly held the upper part of the clipped area using the thumb and forefinger of the left hand.This caused the veins of the lower limbs to fill.With the syringe attached to the 6-gauge needle, the right hand quickly punctured into the vein, and blood was withdrawn at an appropriate speed.Once 2 mL of whole blood was drawn, the needle was removed, and the blood was slowly injected into a 1.5 mL centrifuge tube along the wall.The tube, containing the blood, was placed in a 37℃ warm box for 1 h to promote coagulation.Afterward, centrifugation was performed at 2000 rpm for 10 min.This step aided in collection of the supernatant while avoiding aspiration of impurities.During this process, the serum appeared clear and transparent, either colorless or slightly yellow.Finally, the collected canine serum was stored in portions at -20 °C [27]. Histological analysis After 8 days of treatment, dogs in the dry eye model group and the remaining four groups were euthanized by intraperitoneal injection of an overdose of anesthetics (Ketamine and Xylazine).Periocular tissues, such as corneal tissues and transient membrane glands, were gently separated and collected.These tissues were then washed with PBS and fixed with a 4% paraformaldehyde solution fixative at 4 °C for 24 h.Following fixation, the tissues were gradually dehydrated, embedded in paraffin wax, and sliced into 4 µM sections.Hematoxylin-eosin (H&E) staining and peridynamic acid-schiff (PAS) staining were performed according to the instructions of the reagent vendor (Beyotime).Morphological changes of corneal epithelial cells and transient membrane gland tissues were observed under a light microscope, starting from low magnification and progressing to high magnification.Histopathological observation of the ocular surface was conducted, with 3 sections observed, recorded, and photographed in each eye [36]. Immunohistochemistry staining Eye sections were deparaffinized twice with xylene for 10 min each time.Then, they were deparaffinized with a xylene-ethanol mixture for 5 min, followed by sequential washing with anhydrous ethanol for 5 min, 95% alcohol for 5 min, 85% alcohol for 5 min, 75% alcohol for 5 min, 50% alcohol for 5 min, and distilled water twice for 2 min each time.Next, Tris/EDTA pH 9.0 restoration buffer was used for antigen repair by microwaving the sections for 15-20 min and allowing them to cool at room temperature.After three washes with PBS, the sections were treated with 3% H 2 O 2 for 15 min to eliminate endogenous peroxidase activity, and the tissues were blocked with animal serum.Incubation with PCNA antibody (1:150; immunoway Inc) was carried out overnight at 4 °C.Subsequently, the sections were washed three times with PBS and incubated with horseradish peroxidaselabeled streptavidin working solution for 30 min.After three more washes with PBS, the sections were treated with DAB chromogenic solution for color development and hematoxylin staining solution for nuclear staining.Finally, the samples were dehydrated, sealed with drops of neutral resin, and analyzed under a light microscope [36]. ELISA for inflammatory factors The levels of interleukin-6 (IL-6), interleukin-1β (IL-1β), and tumor necrosis factor-α (TNF-α) in the serum of each group of dogs were determined by double antibody sandwich assay using the 96T ELISA kits for canine IL-6, canine IL-1β, and canine TNF-α, respectively.The assay procedure was performed according to the instructions provided by the reagent supplier (FANKEWEI, Shanghai, China).Five replicate wells were set up in each group of canine sera for the determination of these three inflammatory factors.The absorbance (OD) was measured at 450 nm with an enzyme meter, and the concentrations of IL-6, IL-1β, and TNF-α in the serum of the samples were calculated by a standard curve [26]. Cell scratch experiment Mark a horizontal line on the back of the 6-well plate for observation, inoculate HCECs cells equally into the prepared 6-well plate, waiting for the cells to fully integrate, use a 200 µL tip perpendicular to the marked horizontal line, scrape off the cells with equal widths, and make cell scratches.Excess supernatant should be aspirated, and the cells should be washed twice by adding PBS.The HCECs should be then classified into two groups: the culture group with ADMSC-CMV supernatant and the culture group with ADMSC-OESRARC supernatant.In the first group, 1mL of ADMSC-CMV supernatant and 1mL of DMEM/F12 medium containing 2% serum should be added.In the second group, 1mL of ADMSC-OESRARC supernatant and 1mL of DMEM/F12 medium containing 2% serum were added.Subsequently, the cells were cultured, and the width of the scratch should be photographed and recorded at 12 h and 24 h [27].The culture solution of ADMSC used was as described in 2.2. Giemsa staining In 48-well plates, 250 µL/well of HCECs cell suspension was inoculated.The plates were divided into four groups: normal control group, untreated model group, ADMSC-CMV supernatant culture group, and ADMSC-OES-RARC supernatant culture group.Each group consisted of three replicate wells.Use culture solution as described in 2.9.1.After the cells were adhered to the wall, except for the normal control group, each group was treated with benzalkonium chloride (1 µg/mL per well) for 24 h [39].Following this treatment, the culture medium was changed.In the ADMSC-CMV supernatant culture group, 125 µL of ADMSC-CMV supernatant and 125 µL of DMEM/F12 medium containing 10% serum were added.Similarly, in the ADMSC-OESRARC supernatant culture group, 125 µL of ADMSC-OESRARC supernatant and 125 µL of DMEM/F12 medium containing 10% serum were added.The culture solution of ADMSC used was as described in 2.2.In the untreated model group, 250 µL of normal DMEM/F12 medium containing 10% serum was added, and the cells were cultured for an additional 24 h.Subsequently, the cell supernatant of each group was discarded.The cells were then fixed with 4% paraformaldehyde at room temperature for 20 min, washed three times with PBS, and stained with the configured Giemsa Staining Solution working solution dropwise for 15 min.After three additional washes with PBS, the cells were observed and photographed under a light microscope, following the instructions provided by the Giemsa Staining Kit (Beyotime) [25]. CCK-8 cell proliferation assay The 96-well plate was inoculated with HCECs cell suspension (100 µL/well), divided into four groups, namely the normal control group, untreated model group, ADMSC-CMV supernatant culture group, and ADMSC-OESRARC supernatant culture group.Each group consisted of 5 replicate wells.Use culture solution was described as in 2.9.1.After the cells adhered to the wall, benzalkonium chloride was added to each well (except for the normal control group) at 1 µg/mL for 24 h.Then, the culture medium was changed.For the ADMSC-CMV supernatant culture group, 50 µL of ADMSC-CMV supernatant and 50 µL of DMEM/F12 medium containing 10% serum were added.Similarly, for the ADMSC-OESRARC supernatant culture group, 50 µL of ADMSC-OESRARC supernatant and 50 µL of DMEM/F12 medium containing 10% serum were added.The culture solution of ADMSC used was as described in 2.2.The untreated model group received 100 µL of plain DMEM/F12 medium containing 10% serum.After incubating the cells for an additional 24 h, according to the reagent vendor's instructions (Mishushengwu, Xian, China), 10 µL of CCK-8 solution was added to each well.The cells were incubated for another hour in a cell culture incubator, and the absorbance value at 450 nm was measured using an enzyme marker [40]. Real-time fluorescence quantitative polymerase chain reaction analysis The six-well plates were first inoculated with HCECs cell suspensions (2 mL/well) and then divided into four groups: normal control group, untreated model group, ADMSC-CMV supernatant culture group, and ADMSC-OESRARC supernatant culture group.Use culture solution as described in 2.9.1.The cells were allowed to attach to the wall, and benzalkonium chloride was added to each group (except the normal control group) at a concentration of 1 µg/mL per well for 24 h.After that, the culture medium was changed.In the ADMSC-CMV supernatant culture group, 1 mL of ADMSC-CMV supernatant and 1mL of DMEM/F12 medium containing 10% serum were added.Similarly, in the ADMSC-OESRARC supernatant culture group, 1mL of ADMSC-OESRARC supernatant and 1mL of DMEM/F12 medium containing 10% serum were added.The culture solution of ADMSC used was as described in 2.2.As for the untreated model group, 2mL of plain DMEM/F12 medium containing 10% serum was replaced and the cells were incubated for an additional 24 h, the total RNA was extracted from the HCECs of each group using the TRIzol reagent (Takara, Japan) according to the instructions of reagent vendors.Reverse transcription to cDNA was performed using a reverse transcriptase kit (Thermo Fisher Scientific).quantitative real-time polymerase chain reaction (qRT-PCR) was performed using the CFX96 Real-Time Polymerase Chain Reaction (PCR) System: pre-denaturation at 94 °C for 5 min, denaturation at 94 °C for 30 s, annealing at 58 °C for 30 s, and extension at 70 °C for 30 s. 39 cycles.GAPDH was used as an internal reference.The relative expression of inflammatory factors IL-10, TNF-α, MMP9 and epidermal growth factor EGF was measured using the comparative CT values of qRT-PCR.The primer sequences were as follows: Flow cytometry First, HCECs cell suspension was inoculated into 6-well plates, following the grouping and processing methods outlined in Sect.2.9.5.The cells were then incubated for 24 h.After that, the cell culture solution from each group was aspirated into a suitable centrifuge tube.Next, the adherent cells were washed once with PBS and digested with 0.25% trypsin.The cell culture solution collected earlier was added to the digestion mix, and the cells were gently blown down and transferred to a centrifuge tube.Subsequently, each group of HCECs was stained using the Annexin V-FITC Apoptosis Detection Kit according to the instructions provided by the reagent vendor (Beyotime).After resuspending 100,000 HCECs cells, the suspension was centrifuged, and the supernatant was discarded.Then, 195 µL of Annexin V-FITC conjugate was added to gently resuspend the cells.Next, 5 µL of Annexin V-FITC was added and gently mixed, followed by the addition of 10 µL of propidium iodide (PI) staining solution, also gently mixed.The cells were incubated at room temperature and protected from light for 15 min.Finally, the cells were analyzed using an analytical flow cytometer, and apoptosis was detected in each group.Additionally, In situ FITC/PI double-staining fluorescence assay was also performed on adherent cells to corroborate the results of flow cytometry [42]. Cellular immunofluorescence The cells were inoculated in 96-well plates and grouped according to the treatment described in Sect.2.9.4.Following a 24 h incubation period, the cells were washed with PBS.Subsequently, the cells in each group were fixed using 4% paraformaldehyde at room temperature for 20 min and then washed three times with PBS.To break the membrane, a solution of 0.2-0.5% Triton X-100 dissolved in citrate buffer was added and incubated for 15 min at room temperature.After washing with PBS three more times, each well was blocked with 70 µL of 10% FBS at room temperature for 1 h.Once the FBS was aspirated, 40 µL of primary antibody (PCNA at 1:100 dilution and Ki67 at 1:100 dilution, both from immunoway) was added to each well and incubated overnight at 4 °C.The cells were then washed three times with PBS, and 30 µL of hochest33342 was added to each well.The reaction was carried out at room temperature for 5 min, after which the cells were observed and photographed under a fluorescence microscope [43]. RNA sequencing The cells were divided into ADMSC-CMV group and ADMSC-OESRARC group, and the cell samples were collected and sent to Shanghai Jiayin Biotechnology Ltd. for transcriptome sequencing comparison according to the company's delivery requirements.Gene Ontology (GO) enrichment of potential target gene information and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotation analysis were performed using annotation databases, and pathways with a P value < 0.05 were considered reliable. Statistical analysis Statistical analysis was performed using IBM Corporation's SPSS version 19.0 software (Chicago, IL, USA).All experimental data were examined using one-way analysis of variance (ANOVA) and are presented as mean SD±.Statistical significance was observed in the comparisons if the p-value was less than 0.05.GraphPad Prism software was used to analyze all the data. The viability of ADMSC is enhanced by the overexpression of SPARC To examine the effects of SPARC overexpression, a lentiviral expression vector of pCDH-CMV-SPARC-EF1-copGFP-T2A-Puro was constructed and transfected into 293T cells.In the control group, pCDH-CMV-EF1-copGFP-T2A-Puro was used.Subsequently, viral particles were collected and attacked the ADMSC cell line, resulting in the establishment of steady-transfected green fluorescence-positive cell lines (Fig. 1A).The mRNA levels were evaluated using real-time quantitative PCR and shown more 700-fold up-regulation of SPARC compared to the control group (Fig. 1B).In addition, protein levels were measured using ELISA (Fig. 1C).Furthermore, the overexpression of SPARC exhibited a better promotion effect on ADMSC, as evidenced by the results of Edu staining, cell growth curve, and population doubling.Notably, the cell proliferation rate of ADMSC-OESPARC was significantly faster than that of the ADMSC-CMV group (Fig. 1D, E, F, G).These findings strongly suggest that overexpression of SPARC enhances the activity and proliferation of ADMSC. The efficacy of mesenchymal stem cells in canine dry eye syndrome is enhanced by overexpression of SPARC The canine dry eye model was created using 0.25% benzalkonium chloride ophthalmic spotting.Autologous and allograft recovery around the canine eye was recorded.Ciliary congestion, corneal redness (Fig. 2A, B), corneal damage (Fig. 2C, D), and tear secretion (Fig. 2E) of canine eyes recovered with time in all groups.The slowest recovery was observed in the group of the untreated KCS model, while the fastest recovery was observed in the group of ADMSC-OESPARC.The group of ADMSC-OESPARC showed a more significant effect than that of cyclosporine drops.Repair was basically completed and reached the normal level on the 15th day.These results demonstrate that ADMSC overexpressing SPARC can shorten the repair time of canine corneal injury and alleviate dry eye symptoms. Overexpression of SPARC enhances the beneficial effects of ADMSC in alleviating dry eye through multiple mechanisms The untreated KCS model group showed damaged, disorganized, rough, and even ruptured follicular structure of the transient membrane glands as observed in HE staining.Corneal hyperplasia was also evident, along with disorganized epithelial cell layer, loose and ruptured structure of the cupular cell bundles.In contrast, the ADMSC-OESPARC-treated group exhibited more noticeable improvement compared to the cyclosporine and ADMSC-CMV-treated groups.The treated group showed clear vesicular structures with smooth edges, attenuated corneal hyperplasia, aligned epithelial cell layer with a smoother surface, and a denser stromal layer of the cupular cell bundles(Figs.3A, B).In immunohistochemistry, by PCNA staining, there were fewer brown cells in the untreated KCS model group and an increased brown cells in the ADMSC-OESPARCtreated group, which indicated the best proliferation of cells in this group, and the results of immunohistochemistry were quantified (Fig. 3C, D).Furthermore, the ADMSC-OESPARC treatment group exhibited the most significant decrease in serum levels of IL-6, IL-1β, and TNF-α, indicating a notable suppression of inflammation(Fig.3E).Overall, these results suggest that overexpression of SPARC contributes to the repair of transient membrane glands and cuprocytes, proliferation of corneal epithelial cells, and inhibition of inflammatory response, thereby promoting the efficacy of ADMSC in treating dry eye syndrome. SPARC improves cell viability and anti-inflammatory capacity of HCECs To investigate the relationship between ADMSC-OESPARC on HCECs viability and anti-inflammation, we established an in vitro model of corneal injury by inducing HCECs using 1 µg/mL BAC.We then evaluated the effect of the ADMSC-OESPARC supernatant on cell migration of HCECs.The results showed that the ADMSC-OESPARC supernatant exhibited a higher capability in promoting cell migration of HCECs (Fig. 4A, B).Moreover, Giemsa staining demonstrated that HCEC cell morphology returned to normal after adding the ADMSC-OESPARC supernatant, appearing as smooth and densely packed transverse ellipsoids (Fig. 4C).Additionally, the viability of HCEC cells was markedly improved as detection by CCK8 (Fig. 4D).Furthermore, the expression levels of inflammation-related factors IL-6, TNF-α and MMP9 in HCECs cells significantly decreased after 24 h, whereas the expression of epidermal growth factor (EGF) was up-regulated (Fig. 4E). Effect of SPARC on proliferation and apoptosis of HCECs After BAC induction, flow cytometry detected an increase in late apoptotic and necrotic cells in HCECs.The addition of ADMSC-OESPARC supernatant to culture had the most significant effect on apoptosis(Fig.5A).Furthermore, in situ FITC/PI double-staining fluorescence assay on adherent cells corroborated the results of the flow cytometry, and the fluorescence results were quantified(Fig.5B, C).Moreover, A notable rise in the intracellular PCNA and Ki67 expression in HCECs was observed upon the addition of ADMSC-OESPARC supernatant culture and the cellular immunofluorescence results were quantified (Fig. 5D, E).The results above indicate that ADMSC overexpressing SPARC influence the migration and differentiation of HCECs by enhancing Transcriptome analysis of ADMSC reveals the differentially expressed genes and pathways influenced by SPARC In order to further explore the target genes and downstream signaling pathways regulated by SPARC, transcriptome sequencing was performed on ADMSC-OESPARC and ADMSC-CMV.The sequencing results showed that 60 genes were up-regulated and 66 genes were down-regulated in the ADMSC-OESPARC group (Fig. 6A).The heat map showed the representative genes positively and negatively regulated by SPARC.The negatively regulated genes included CD40, CCL20, COL8A2 and IGFBP2, and the positively regulated genes included IL18R1, VCAN and SRGN (Fig. 6B).KEGG analysis of these up-regulated and down-regulated differentially expressed genes revealed that they were enriched in KEGG pathways, including MAPK, VEGF, TNF and PI3K-Akt signaling pathways (Fig. 6C, D).The enriched GO pathways mainly included cell adhesion, cell response to lipid, positive regulation of glial cell proliferation, and regulation of angiogenesis, as revealed by the GO analysis performed on the up-regulated and down-regulated differentially expressed genes (Fig. 6E, F). Discussion The key to the pathogenesis of dry eye lies in the change in the stability of the tear film on the corneal surface.This change can lead to corneal epithelial defects, which not only damage the normal structure and function of the corneal epithelium but also potentially result in the partial or complete loss of the corneal epithelial cell layer.Consequently, this can impede or delay the repair process, creating a detrimental cycle that complicates the treatment of dry eye [44].Impaired stability of corneal cells increases the osmotic pressure of tears, thereby activating the mitogen-activated protein kinase (MAPK) and nuclear factor-κB (NF-κB) signaling pathways in the corneal epithelium [44].This activation triggers the release of various inflammatory mediators, including interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), and matrix metalloproteinases (MMPs), especially MMP-9.MMP-9 can disrupt the tight junction protein complex, leading to corneal epithelial cell shedding, filamentous keratitis, and an increased likelihood of developing moderate to severe dry eye [45].ADMSC have gained attention for their therapeutic potential in treating chronic ophthalmic conditions due to their regenerative, repair, and immunomodulatory capabilities.When ADMSC are used in dry eye dogs, they can migrate to the injured corneal site and enhance the proliferation and differentiation of corneal cells [46].However, despite their promising attributes, single mesenchymal stem cell therapy has limitations, such as poor cell activity and low cell survival rates.SPARC plays a crucial role in the regulation of the extracellular matrix.It can affect the stability of cellmatrix adhesion of the cytoskeleton and regulate cell adhesion through interaction with cell membrane receptors of mesenchymal stem cells, thereby enhancing the stability of mesenchymal stem cells in a damaged microenvironment [47].Research indicates that SPARC can modulate the activity of cell proliferation-related signaling pathways, such as the PI3K/Akt and ERK/MAPK pathways.Activating these signaling pathways aids in boosting the proliferation capacity of mesenchymal stem cells, ultimately enhancing their ability to repair damaged corneal epithelial cells [48].Furthermore, the expression of SPARC undergoes regulation during corneal repair, which is closely intertwined with the pathogenesis and progression of eye disorders like dry eye [49].Following a corneal injury, corneal cells and adjacent stromal fibroblasts become activated and migrate to the affected area to engage in the reparative process.These new stromal cells, rich in endoplasmic reticulum, have the capacity to synthesize cytoskeleton components, adhesion molecules, and various collagen types [50].By interacting with corneal stromal cells, SPARC significantly influences various processes, including collagen fiber contraction, regulation of cell-matrix forces, modulation of cell morphology, and ultimately facilitating the repair of corneal injuries [51].To validate this concept, an ADMSC cell line that overexpresses SPARC was engineered in this study.Through evaluating the cell line's growth and proliferation abilities, it was discerned that SPARC can enhance the activity, proliferation rate, and growth potential of ADMSC in vitro.Analysis of transcriptome sequencing data further corroborated that SPARC contributes to promoting lipid synthesis, vascular development, cell adhesion, collagen fiber regulation, anti-inflammatory responses, and other essential biological processes within ADMSC.This substantiates the notion that SPARC augments the healing and growth capabilities of ADMSC in the context of corneal epithelium restoration. To investigate the therapeutic potential of ADMSC in treating dry eye disease, this study utilized Benzalkonium Chloride (BAC) to induce a dry eye disease model in dogs and conducted live experiments.Benzalkonium chloride, a common preservative found in eye drops, can lead to dry eye symptoms due to its detrimental effects on the connexin of the corneal epithelium and the lipid layer of the tear film upon prolonged excessive usage [52].Recent research has increasingly used BAC to establish animal models of dry eye disease.Notably, studies have demonstrated that applying a 0.25% BAC solution to the eyes of rabbits for 14 days resulted in reduced tear secretion and corneal sodium fluorescein staining by the third day, and the appearance of cup cells by the seventh day [29].Building upon these findings, a dry eye model was induced in mice by utilizing a 0.1% BAC solution.By the 7th day, mice exhibited corneal inflammation and heightened apoptosis of corneal epithelial cells [53].In this investigation, a 0.25% BAC solution was employed to create a moderate to severe dry eye syndrome model in dogs.As the blood of dogs from each group needed to be collected for serological analysis post-treatment, each dog in the study was considered an independent experimental sample, and no internal control model was established.Using established criteria for grading dry eye severity, model dogs displaying moderate to severe clinical symptoms were selected to evaluate the therapeutic potential of ADMSC overexpressing SPARC in treating dry eye.Given the potential complications associated with ADMSC therapy, the dosage of cells administered is crucial to ensure treatment safety and efficacy.Typically, the recommended dosage range for administering ADMSC in ophthalmic diseases in animal models is between 0.5 × 10 5 to 1 × 10 6 cells per kilogram of body weight, for either systemic or local delivery [30,31].Despite the absence of consensus on optimal dosages, empirical evidence suggests a dose-dependent immunosuppressive effect of ADMSC within a defined range, beyond which higher doses do not confer additional therapeutic benefits.This study determined that administering 1 × 10 5 ADMSC (equivalent to 0.5 × 10 5 cells/kg) was both safe and efficacious in treating dry eye.Furthermore, the delivery route of therapeutic agents significantly impacts drug absorption and utilization efficiency.While intravenous injection is commonly used in mesenchymal stem cell studies, its efficacy may be limited when treating eye diseases, potentially leading to pulmonary embolism risk [54].Alternative methods for ocular administration in experimental animals include eye drops, subconjunctival and intravitreal injections [55].Of these, eye-drop administration is considered safer and more convenient, albeit with reduced ocular surface residence time and efficiency of drug absorption, so we adopt the means of repeated administration to make the treatment effect more significant.In this research, a cell suspension overexpressing SPARC was applied to the affected dog's ocular surface.Treatment efficacy was assessed based on the clinical rating standard for dry eye disease, pathological biopsy examinations, and serum inflammation indicators.Results indicated that SPARC modification enhanced the therapeutic effects of ADMSC by expediting corneal swelling recovery, promoting nictitating gland, corneal, and conjunctival tissue repair, enhancing tear secretion restoration, reducing inflammation in dry eye dogs, and accelerating epithelial cell proliferation in corneal tissue.This suggests that SPARC-modified ADMSC may enhance local administration efficacy and exhibit improved therapeutic benefits for dry eye. We aimed to further ascertain the enhancing impact of SPARC on the efficacy of ADMSC at the cellular level, given that the destabilization of corneal epithelial cells stands as the focal point of dry eye pathogenesis.To induce injury in human corneal epithelium cells (HCECs) in vitro, we utilized a 1 µg/mL BAC solution, following the investigation conducted by Jillian F. et al. into the effects of benzalkonium chloride concentrations ranging from 0.2 to 200 µg/mL on the lipid set of HCECs [39].Subsequently, a non-contact two-dimensional co-culture model was established by adding ADMSC-OESPARC supernatant to the injured corneal epithelial cell lines for analysis of the therapeutic potential of ADMSC-OESPARC [56].The outcomes revealed a restoration of HCECs morphology towards a normal transverse oval shape and a remarkable enhancement in cell viability.Furthermore, following a 24-hour co-culture period, there was a notable decrease in the expression levels of inflammatory mediators IL-6, TNF-α, and MMP-9 released by HCECs, coupled with an up-regulation in the expression level of epidermal growth factor EGF.It was observed that SPARC played a crucial role in boosting the reparative effects of ADMSC on injured corneal epithelial cells by fostering cell proliferation and impeding cell apoptosis.This investigation contributes vital insights and robust evidence for the management of dry eye disease, underscoring its significance in clinical interventions for dry eye disease in veterinary practice.Moreover, drawing on the groundwork of animal studies, this research is anticipated to offer valuable insights for the exploration of related human conditions.Accordingly, our data encourage the need for a prospectively randomized clinical trial to investigate the efficacy and safety of such a combined regimen for patients with KCS. In our study, we have addressed certain questions regarding the effects of SPARC-modified ADMSC.However, long-term effects of this treatment remain unclear.Furthermore, the optimal dose and frequency of ADMSC administration require further investigation, especially considering the large size of dogs and the limited sample size in our study.The exact mechanism underlying the potential immunomodulatory effects of SPARC-modified ADMSC also warrants further exploration.Additionally, the clinical application of this therapy presents challenges that need to be considered in future research endeavors.These challenges include determining how to manage the cost associated with mass expansion of mesenchymal stem cells, preparing cellular agents, and developing more convenient drug delivery technology. Conclusions This study demonstrated that SPARC-modified ADMSC showed superior efficacy compared to either ADMSC or cyclosporine eye drops alone in promoting corneal healing, up-regulating tear secretion, and suppressing periocular inflammation in a canine model of dry eye induced with 0.25% BAC.The cellular supernatant of SPARC-modified ADMSC not only strengthened the cellular viability and anti-inflammatory capacity of HCECs but also proved to be the most effective in rescuing 1 µg/ mL BAC-induced damage to HCECs.Furthermore, the co-sequencing results revealed that SPARC played a crucial role in promoting the repair of corneal epithelial cells and regulated the production of inflammatory mediators by increasing the in vitro viability, migration and proliferation, immunosuppression, regulation of neovascularization, and anti-scarring of ADMSC.This evidence suggests the potential future clinical application of SPARC in combination with ADMSC as cell therapy. This study offers important insights for the development of new approaches in treating dry eye syndrome. Fig. 2 Fig. 2 The efficacy of ADMSC in canine dry eye disease is enhanced by overexpression of SPARC.(A) Observed variations in ocular surface inflammation in canines.(B) Ocular surface inflammation index score.(C) Corneal sodium fluorescein staining changes recorded.(D) Corneal sodium fluorescein staining score.(E) Record of basal tear secretion Fig. 3 Fig. 3 Overexpression of SPARC enhances the beneficial effects of ADMSC in alleviating dry eye through multiple mechanisms.(A) HE staining is used to observe the structure of transient membrane glands.(B) HE staining is used to examine the structure of corneal epithelial cells.(C) Immunohistochemistry for PCNA expression observation.(D) Quantitative analysis of immunohistochemical results.(E) ELISA is used to detect inflammatory factors.ns: Not significant; * : P < 0.05 ; : P < 0.01 ; * : P < 0.001 Fig. 4 Fig. 4 SPARC improves cell viability and anti-inflammatory capacity of HCECs.(A) The HCECs cell scratch assay detects the cell migration ability of each group.(B) Quantitative analysis of cell migration assay.(C) Giemsa staining was used to observe the morphology of HCECs in all groups.(D) CCK8 detection of cell viability in each group.(E) The mRNA expression of IL-6, TNF-α, MMP9 and EGF in each group.ns: Not significant; * : P < 0.05 ; : P < 0.01 ; * : P < 0.001	2024-07-04T05:07:58.877Z	2024-07-02T00:00:00.000	{ "year": 2024, "sha1": "7e95b5814e141f7e3eff85c3acfca1a5b1d6a3f9", "oa_license": "CCBY", "oa_url": "https://doi.org/10.1186/s13287-024-03815-z", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "2197c1fe0fff5d9fc12142d59b59c1b6c4f8634f", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
119159549	pes2o/s2orc	v3-fos-license	Conformal polynomial parameterizations The current paper discusses some new results about conformal polynomic surface parameterizations. A new theorem is proved: Given a conformal polynomic surface parameterization of any degree it must be harmonic on each component. As a first geometrical application, every surface that admits a conformal polynomic parameterization must be a minimal surface. This is not the case for rational conformal polynomic parameterizations, where the conformal condition does not imply that components must be harmonic. Finally, a new general theorem is established for conformal polynomic parameterizations of m-dimensional hypersurfaces, m>2, in R^n, with n>m: The only conformal polynomic parameterizations of a m-dimensional hypersurfaces, in R^n, with m>2 and n>=m, must be formed by lineal polynomials, i.e. the parameter must be a rotation, scale transformation, reflection or translation of the usual cartesian framework. As a first geometrical application, every surface that admits a conformal polynomic parameterization must be a minimal surface. This is not the case for rational conformal polynomic parameterizations, where the conformal condition does not imply that components must be harmonic. Finally, a new general theorem is established for conformal polynomic parameterizations of m-dimensional hypersurfaces, m > 2, in R n , with n > m: The only conformal polynomic parameterizations of a m-dimensional hypersurfaces, in R n , with m > 2 and n ≥ m, must be formed by lineal polynomials, i.e. the parameter must be a rotation, scale transformation, reflection or translation of the usual cartesian framework. 1 Conformal polynomial parameterizations 1.1 Harmonic and homogeneous polynomials Definition 1. A surface parameterization is conformal if satisfies the following condition: X x ·X x −X y ·X y = 0 X x ·X y = 0 (1.1) In polar coordinates the condition is: Definition 2. A polynomial p is said to be homogeneous when all the monomial components has the same degree. In other words, p is a homogeneous polynomial of degree k in R n if it has the following form: where summation contains all combinations that satisfy n j=1 α i j = k ∀i. The set of n variables homogeneous polynomials will be denoted by P(R n ) and the set of homogeneous polynomials of degree i by P i (R n ). Definition 3. A polynomial p is said to be harmonic when its laplacian is null, i.e. ∆p = 0. H(R n ) will denote the harmonic homogeneous polynomials set in R n and H i (R n ) symbolize the harmonic homogeneous polynomial set of degree i. The following known result, based on a more general one proved by Ernst Fischer in 1917 [6], will be used: Theorem 1. Every homogeneous polynomial could be uniquely decomposed as a sum of harmonic homogeneous polynomials multiplied by r 2 powers. More explicitly, every m degree homogeneous polynomial p ∈ P m (R n ), could be decomposed as: ([ ] is the integer part operator), and r 2 is the square of the position vector module in R n , r 2 = x 2 1 + · · · + x 2 n = \|x\| 2 , and every h i are harmonic homogeneous polynomials of degree i, h i ∈ H i (R n ). The proof of this theorem could be seen in [1], theorem 5.7. In terms of polynomial spaces, the homogeneous polynomial space admits the decomposition in harmonic polynomial spaces, given by: In the cited reference, [1], could be seen a proof of the following proposition about the harmonic polynomial space dimension: In order to study conformal surface parameterizations only harmonic polynomials of two variables, H m (R 2 ) will be used. Following the above proposition, the harmonic polynomial base of any degree is always compound by two elements. Remark 1. The harmonic two variables k degree polynomial space, H k (R 2 ), could be expressed as the real and imaginary part of z k , where z ∈ C. A base of the space H k (R 2 ) is {Re(z k ), Im(z k )}. In other words, every H k (R 2 ) element is a lineal combination of {z m , z m }. One harmonic homogeneous polynomial base in polar coordinates is: this is the known Fourier base and the decomposition exposed in the theorem 1, applied in the two variable case, is the Fourier series expansion of any homogeneous polynomial. The two elements of the standard basis of H m (R 2 ) will be denoted by Remark 2. The harmonic polynomial decomposition stated by theorem 1 and a vector coefficient notation will be used. The next example tries to clarify the notation. The standard basis of H i (R 2 ), i = 1, 2, 3, are the next harmonic homogeneous polynomial pairs: The Enneper minimal surface has polynomic components and also is conformal. This surface could be expressed in terms of vector coefficients multiplied by harmonic base elements as: where: The notation for the angular component of the k degree Fourier basis elements will be:f k =v k sin kθ +¯k cos kθ where the radial factor, r k , is deliberately eliminated. The next definition will also be used: The next relations are consequences of the Fourier basis orthogonality properties in the unit circle S 1 , (r = 1; 0 ≤ θ ≤ 2π): whereō k andq k are the vector coefficients ofḡ k , the angular component of another harmonic polynomial. A conformal polynomial parameterization theorem Theorem 3. Every conformal polynomial surface parameterization, embedded in R n , must be harmonic. The general form of any conformal polynomial parameterization of k degree in polar coordinates is: or using cartesian coordinates: Also the vector coefficients of maximum, j = k, and minimum degree, j = 1, must satisfy: Proof. The proof is decomposed in the following steps: 1. Take a polynomial surface parameterization, in R n , of maximum degree k:X (x, y) = (X 1 (x, y), . . . , X n (x, y)) 2. The polynomial parameterization splits into the sum of homogeneous components of degrees 1, . . . , k. The constant terms are neglected because the conformal condition is invariant under surface translations. Using the introduced vector notation, the polynomial surface parameterization takes the form: 3. The decomposition theorem 1 is applied to each homogeneous component:P and vectorsf i ,ḡ i , . . . ,h i symbolize the angular component of the i degree homogeneous polynomials and following the previous polar notation:f The k degree polynomial parameterization takes the form: The maximum order harmonic terms,f k ,f k−1 , . . . will be called principal harmonic components of the harmonic decomposition. 4. The tangent vectors are expressed in polar coordinates, the rX r components are: This equalities are replaced on the first conformal parameterization condition (1.2): The last term groups odd r powers because it will be null when the equation is integrated on the unit circle S 1 , by the harmonic polynomial orthogonality properties (1.5). 5. Coefficients of different r powers must be null simultaneously, because conformal condition must be satified in all the space. The following relations are obtained: [Odd r powers] = 0 6. Now each equation is integrated on the unit circle. Using the cited Fourier basis orthogonality properties, (1.5), the last term, corresponding to products of principal harmonic components, have identical coefficients The equations could be simplified and take the form: . . . There are only two types of products of the same degree: • Terms corresponding to square powers of harmonic components, like: • Cross-products of different degree harmonic components, for example: Using the orthogonality properties (1.5), the above equations are sim-plified into: The second type of terms, the cross-products, are removed gradually. Each equation, that represents the coefficient of a different r power, make null the square of the terms that appears on the cross-terms of the next r power. On the first equation, the square elements must be null, because all the coefficients are positive and could be deduced: In the second equation the cross-product terms vanish because it contains the product of the harmonic components g k−2 : If we remove this cross term, the new equation cancel the harmonic coefficients of the next lower degree: 8. This procedure is iterated over all the equations, the only not null resulting terms are the square terms of the principal harmonic components. The coefficients of the squares of non principal harmonic components are positive (it takes the form k 2 − (k − i) 2 , with i < k and k > 1) so all the quadratic terms must be null simultaneously: Therefore, all non principal harmonic components are null. The polynomial parameterization could only contain principal harmonic components. In other words, the function components of the polynomial parameterization must be harmonic. 9. Now could be applied the conformal parameterization condition (1.2) to the harmonic polynomial parameterization and then r powers could be grouped obtaining additional conditions for vector coefficients. The more simple conditions, for the higher, j = k, and lower vector coefficients, j = 1, are: Remark 3. The above proof only uses the first conformal parameterization condition (1.2):X x ·X x −X y ·X y = 0 It could be thought that the second condition: X x ·X y = 0 imposes additional restrictions. In general, for non polynomial parameterizations, that is true. In the polynomial case it will be shown that this condition is superfluous. The conformal polynomial parameterization could be written on the complex plane as: The surface tangent vectors are given by: The first condition on the complex plane is: This condition takes the form:X z ·X z = 0 The second conformal parameterization condition on the complex plane is:X If the parameterization X is harmonic and polynomial, it could be expressed as a polynomial in z variable, see remark 1, i.e., it must be an holomorphic function. Whenz derivatives are neglected it could be seen that the two conditions are equivalent. Geometric applications 2.1 Minimal surfaces We use the next notation for the elements of the metric or the first fundamental form: Using this notation, a surface parameterization is conformal when: The elements of the second fundamental form are: The mean curvature expression, see for example [4], as a function of the first and second fundamental forms is: The following geometric result is obtained from the theorem 3 of the previous section. Corollary 4. Every Riemannian surface M in R n that admits a conformal polynomial parameterization must be a minimal surface. Proof. LetX be a conformal parameterization. The theorem 3 states that every conformal polynomial parameterization must be also harmonic, ∆X = 0 In the conformal parameterization case, because E = G, F = 0, the mean curvature is given by: The fact that the parameterization has to be harmonic, in all of his components, implies that surface mean curvature must be null, i.e. the surface must be minimal. Remark 4. The use of conformal parameters is very common in the resolution of physical and engineering problems because many describing differential equations are simplified making use of this kind of surface coordinates (the local existence of conformal coordinates is guaranteed as can be seen in [4] or [5]). In addition to this, many times the solutions are approximated by a Taylor polynomial, or any other polynomial series expansion, around a point. Using the above result, this kind of aproximmations are really minimal surface approximations. In the next section it will be seen that this condition is still more restrictive for conformal polynomic hypersurfaces. Conformal spinorial surface representation The Weierstrass-Enneper surface representation is used to generate minimal surface conformal parameter based on two complex functions. This idea has been extended to obtain corformal parameters of any kind of surface, not only the minimal ones, in R 3 and R 4 . See [13], [7], [10], [9], [11], [8] for a detailed description. The spinorial differential equations used to obtain the components of the conformal parameter, see [13], are: where D corresponds to the complex bidimensional Dirac operator and A is a real scalar potential: As noted by [13] this equations corresponds to the stationary Dirac equation in presence of a external scalar electromagnetic field. The theorem 3 implies that there are no polynomial solution for the above equation with non null potentials. In other words, there are non polynomial spinors (the components of the stationary wave function) when potential A is not null. Homogeneous system, A = 0, corresponds to spinorial representation of minimal surfaces. A rational conformal counterexample The theorem 3 establish that every conformal polynomial parameterization of a embedded surface in R n must be harmonic and, by the corollary 4, it also must be a minimal surface. That is not the case for conformal rational polynomial parameterizations, where the components are quotients of polynomials. One counterexample could be found using a conformal transformation of a known conformal polynomial parameterization, for example the Enneper minimal surface. The conformal transformation group is the transformation group that conformally changes R n . For n > 2, as the Liouville theorem states, this group is composed by translations, scale transformations, rotations and special conformal group transformations, SCG(R n ). The special conformal group, SCG(R n ), is a subgroup which elements could be expressed as the composition of a inversion, R, a translation by a vectorā, called T (ā) and a new radix inversion R. In other words, for every S ∈ SCG(R n ) exist a vectorā that satisfies S = R · T (ā) · R. This kind of conformal transformation could be applied to a conformal parameterization of the Enneper minimal surface to obtain a new conformal parameterization of a different surface in R n . The new conformal surface is not necessarily minimal although the original surface is minimal because the mean curvature H is not a conformal invariant (not as the Willmore integrand, (H 2 − K)dσ, where dσ is the area differential, that is conformally invariant). A lot of counterexamples of conformal rational polynomial surface parameterizations could be obtained with not everywhere null curvature, i.e., the new surfaces are not minimal and also not harmonic. There are also non Willmore surface examples obtained using interesting spinorial technics that will not be included here. A general theorem for hypersurfaces The idea of the theorem 3 could be generalized to m-dimensional hypersurfaces in R n . It will be seen that there are rigidity conditions, as restrictive as the established by the Liouville theorem. The classical Liouville theorem states that every conformal transformation of a space region in R n , with n > 2, could be expressed as a composition of some of the following operations: inversions, translations, rotations y scale transformations. For a proof see for example [2] or [12], vol 3. In fact, it will be shown that the only conformal polynomial parameterizations of a hypersurface must be composed lineal polynomials. In other words, every conformal polynomial parameterization of a m-dimensional hypersurface, embedded in R n , is essentially a rotation, translation or scale transformation of the cartesian framework. Theorem 5. Every conformal polynomic parameterization of a m-dimensional surface, with m > 2, embedded in R n , must be lineal, i.e., it must be a hyperplane. In other words, the surface parameterization must be a lineal conformal transformation (rotations, translations or scale transformations) of the mdimensional cartesian framework. Proof. Letψ be a conformal polynomial parameterization of a m-dimensional surface embedded in R n and letφ be a conformal polynomial parameterization of a bidimensional surface embedded in R n . The two conformal polynomial parameterizations are: The composition of both maps,X : R 2 → R n , must be a conformal map too and it is a conformal polynomial parameterization of a bidimensional surface in R n . The theorem 3 establish that this parameterization must also be harmonic. Thus every component, X i , of the parameterization: , y), . . . , φ m (x, y)) must be harmonic, ∆X i = 0. The laplacian of each component could be calculated explicitly: The term ∆φ j could be removed because the components are conformal and polynomic and has to be harmonic. Finally, the laplacian of the parameter components X i is: where Hess(ψ i ) is the hessian matrix of the ψ parameterization i component. The previous relation must be true for every conformal parameterization φ, of a surface in R m , used on the compositionX =ψ(φ(x, y)). The simpler polynomial parameterization is a lineal one: φ(x, y) =λx +βy In order to be conformal it must satisfy: The relation (3.6) must be satisfied at every point p ∈ R m : ∆X i p =λ · A ·λ +β · A ·β = 0 (3.7) with A ≡ Hess(ψ i ) p , where A is a symmetric matrix because it symbolize the Hessian matrix evaluated at the point p. The later equation is equivalent to the projection of a bilinear form represented by the symmetric matrix A into the plane formed by the vectorsλ,β. The relation must be true for every conformal parameterization φ, the pair of vectors {λ,β} could be chosen to be A matrix eigenvectors, {v 1 ,v 2 }. The only requirement for the vectorsλ,β is that they have to be orthogonal and of identical size. For examble, it could be taken unitary. This condition is fulfilled by the matrix A eigenvectors because A is real and symmetric. The spectral theorem for finite spaces states that every real symmetric matrix could be diagonalized in a orthogonal basis and the eigenvalues must be real numbers. The same reasoning could be applied to the other m eigenvectors of the A matrix: λ i + λ j = 0 ∀i = j i, j = 1, . . . , m This lineal equation system has only null solution for m > 2. In other words, the matrix A must be null, or equivalently, Hess(ψ i ) = 0 at every point p. All the second order derivatives and the second order cross derivatives of the parameterization components must be null and thus the conformal polynomial parameterization must be lineal. Remark 5. The above result is coherent with the Liouville theorem, when the dimension values m, n are the same. In the case n = m, the only conformal polynomial transformations, from R m to R m , allowed by the Liouville theorem are the lineal ones. This lineal transformations corresponds to the composition of rotations, scale transformations and translations. The special conformal subgroup transformations could not be used because it contains radix inversions that are not polynomial transformations.	2012-05-25T14:29:50.000Z	2012-05-25T00:00:00.000	{ "year": 2012, "sha1": "619f3de9b4034d02ae02cd69234c6462cb5a028a", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "619f3de9b4034d02ae02cd69234c6462cb5a028a", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics" ] }
239528260	pes2o/s2orc	v3-fos-license	IgA Vasculitis with Nephritis in Adults: Histological and Clinical Assessment Patients with IgA vasculitis (IgAV), an immune complex-mediated disease, may exhibit kidney involvement—IgAV with nephritis (IgAVN). The kidney-biopsy histopathologic features of IgAVN are similar to those of IgA nephropathy, but little is known about histopathologic disease severity based on the interval between purpura onset and diagnostic kidney biopsy. We assessed kidney histopathology and clinical and laboratory data in a cohort of adult patients with IgAVN (n = 110). The cases were grouped based on the interval between the onset of purpura and kidney biopsy: Group 1 (G1, <1 month, n = 14), Group 2 (G2, 1–6 months, n = 58), and Group 3 (G3, >6 months, n = 38). Glomerular leukocytes were more common in G1 than in the other groups (p = 0.0008). The proportion of neutrophils among peripheral-blood leukocytes was the highest in the patients biopsied within a month after onset of purpura (G1: 71 ± 8%). In the patients with an interval >6 months, the neutrophil proportion was lower, 60%. Moreover, the glomerular mesangial proliferation score correlated with the serum total IgA concentration (p = 0.0056). In conclusion, IgAVN patients biopsied <1 month from purpura onset showed an elevated percentage of blood neutrophils and glomerular leukocytes, consistent with an acute-onset inflammatory reaction. In all IgAVN patients, the mesangial proliferation score correlated with the serum IgA level. Introduction IgA vasculitis (IgAV), formerly known as Henoch-Schönlein purpura, is a systemic immune complex-mediated, small-vessel leukocytoclastic vasculitis. It is characterized by nonthrombocytopenic palpable purpura and/or arthritis, and abdominal pain [1]. IgAV, the most common vasculitis in children, is often a self-limiting and benign disease that spontaneously resolves. IgAV in adults has a more severe course and poor outcome due to the high frequency of glomerulonephritis, i.e., IgAVN-the most serious complication of this vasculitis. However, there is limited information about the histopathologic disease severity as related to disease onset. In this study, we assessed the kidney histopathology and clinical and laboratory data in a large cohort of adult patients with IgAVN whose diagnostic kidney biopsies had been performed at different intervals after the onset of purpura. IgAVN Patients This is a retrospective study of 110 adult patients with IgAVN. All patients underwent a diagnostic kidney biopsy between 2002 and 2013 at the Department of Nephrology, Huashan Hospital, Fudan University, Shanghai, China. The diagnosis of IgAVN was based on the documented hematuria and proteinuria associated with a characteristic purpuric eruption, or abdominal or joint pain. IgA as the predominant mesangial immunoglobulin as per immunofluorescence microscopy in kidney biopsy specimens was required for inclusion in the study. We excluded patients whose biopsies contained <8 glomeruli, a number considered inadequate for appropriate histopathologic scoring (12), as detailed below. The appearance of purpura defined the onset of IgAV. The 110 patients were divided into three groups based on the interval between purpura onset and diagnostic kidney biopsy: Group 1, <1 month (n = 14); Group 2, 1-6 months (n = 58); and Group 3, >6 months (n = 38; mostly >1 year, with the longest interval 13 years). All biopsies were performed for subjects who exhibited proteinuria in an outpatient setting (>0.5 g/24 h or urinary albumin/creatinine ratio >300 mg/g). The clinical features, laboratory data from the time of biopsy, and histopathologic findings in the biopsy specimens were retrospectively compiled for each group. However, no data from the outpatient follow-up visits since the onset of purpura or after kidney biopsy were available for this study. No personal identification information was collected. The study was approved by the ethics board of Huashan Hospital, Fudan University, Shanghai, China. Clinical and Demographic Data All clinical, laboratory, and demographic data were collected at the time of the kidney biopsy. The demographic data included age, gender, and comorbidities such as hypertension, diabetes mellitus, and cardiovascular disease. The characteristics of IgAVN included skin rash, gastrointestinal and joint manifestations, and kidney involvement. Proteinuria was evaluated by a 24-h urine measurement, hematuria was defined as 22 or more red blood cells (RBC) per microliter of urine (microscopic), or visible hematuria (macroscopic). The serum albumin, creatinine, urea nitrogen, uric acid, complement C3, and total IgA levels were measured at the time of kidney biopsy in the central clinical laboratory of Huashan Hospital. The number of urinary RBC had been determined by using a Sysmex UF-1000i analyzer (Siemens, Germany). Peripheral blood cell profiling was performed using a Sysmex xn-2000 analyzer (Siemens, Germany) and expressed as the total number of leukocytes and relative proportions of neutrophils and eosinophils (as % of total leukocytes). Kidney Pathology Two pathologists examined and graded the histopathological changes. To evaluate the glomerular mesangial-cell proliferation, the cellularity of each glomerulus was graded as per the Oxford classification [12] (<4 mesangial cells/mesangial area = 0; 4-5 mesangial cells/mesangial area = 1; 6-7 mesangial cells/mesangial area = 2; >8 mesangial cells/mesangial area = 3), and a mean mesangial score was calculated for each biopsy. Mesangial Score: sum of grades divided by number of glomeruli (excluding globally sclerotic glomeruli). Crescents (%): number of glomeruli with crescents divided by number of glomeruli × 100. Leukocyte infiltration of glomeruli was considered significant when five or more polymorphonuclear and mononuclear cells per glomerulus were observed using the periodic-acid Schiff-stained tissue sections [11,12] (Figure 1). Statistical Analyses Normally distributed variables are expressed as the mean ± standard deviation (SD and were compared using one-way analysis of variance (ANOVA) or Student's t-test Nonnormally distributed variables are expressed as the median with interquartile rang and were compared using the rank sum test. Categorical variables are expressed as per centages and compared using Pearson's chi-square test or Fisher's exact test. All tests wer two-tailed, and statistical significance was defined as p < 0.05. The SPSS statistical softwar program (version 15.0, SPSS Inc., Chicago, IL, USA) was used for all analyses. Baseline Clinical Data at the Time of Kidney Biopsy IgAVN patients (n = 110) in this study had a mean age of 36.5 ± 16.0 years at the tim of kidney biopsy and consisted of 50 males and 60 females ( Table 1). All patients presented with cutaneous purpura on at least one occasion; purpura was associated with arthralgi in 30 cases (26%) and with arthralgia and abdominal pain in 31 cases (27%). At the time o kidney biopsy, proteinuria ≥0.30 g/24 h was detected in 105 patients (92%) and 73 patient had proteinuria ≥1 g/24 h (64%). Five patients with proteinuria ≥0.5 g/24 h originally meas ured in the outpatient clinic had proteinuria <0.30 g/24 h later on admission to the hospita for kidney biopsy, likely due to prior treatment with an angiotensin-converting enzym inhibitor (ACEi) and/or angiotensin receptor blocker (ARB). Statistical Analyses Normally distributed variables are expressed as the mean ± standard deviation (SD) and were compared using one-way analysis of variance (ANOVA) or Student's t-test. Nonnormally distributed variables are expressed as the median with interquartile range and were compared using the rank sum test. Categorical variables are expressed as percentages and compared using Pearson's chi-square test or Fisher's exact test. All tests were two-tailed, and statistical significance was defined as p < 0.05. The SPSS statistical software program (version 15.0, SPSS Inc., Chicago, IL, USA) was used for all analyses. Baseline Clinical Data at the Time of Kidney Biopsy IgAVN patients (n = 110) in this study had a mean age of 36.5 ± 16.0 years at the time of kidney biopsy and consisted of 50 males and 60 females ( Table 1). All patients presented with cutaneous purpura on at least one occasion; purpura was associated with arthralgia in 30 cases (26%) and with arthralgia and abdominal pain in 31 cases (27%). At the time of kidney biopsy, proteinuria ≥0.30 g/24 h was detected in 105 patients (92%) and 73 patients had proteinuria ≥1 g/24 h (64%). Five patients with proteinuria ≥0.5 g/24 h originally measured in the outpatient clinic had proteinuria <0.30 g/24 h later on admission to the hospital for kidney biopsy, likely due to prior treatment with an angiotensin-converting enzyme inhibitor (ACEi) and/or angiotensin receptor blocker (ARB). Histopathology of Kidney Biopsy Specimens with Different Intervals between Purpura Onset and Diagnostic Kidney Biopsy We next assessed whether the kidney pathology findings differed based on the interval between purpura onset and diagnostic kidney biopsy. The MEST-C scores (12) were calculated (Table 3). Using ANOVA and Student's t-test, the only significant difference between the groups was for M1 between Group 2 and Group 3 (p = 0.006). Furthermore, glomerular leukocytes were more common in Group 1 (57%) compared to Group 2 and Group 3 (p = 0.0008) ( Table 3). Thus, IgAVN patients with kidney biopsy less than one month after purpura onset more frequently had leukocytes in the glomeruli. Furthermore, M1 was more common in Group 3 than in Group 2. Association of Serum Total IgA Concentration and Mesangial Proliferation The serum total IgA concentration positively correlated with the glomerular mesangialproliferation score (p = 0.0056) (Figure 2). Circulating levels of Gd-IgA1 and Gd-IgA1-specific IgG autoantibodies are elevated in patients with IgAVN but not in patients with IgAV [29], supporting the hypothesis that IgAVN and IgAN share pathogenetic components. IgAVN patients have the onset of disease defined by purpura, with kidney involvement developing with 4-6 weeks later [5,13,17]. However, there is a limited information about histopathologic disease severity in relation to disease onset. In this study of 110 adult patients with IgAVN, we assessed histopathologic disease severity based on interval between purpura onset and diagnostic kidney biopsy and correlated the findings with clinical and laboratory data. IgAVN patients biopsied <1 month since the onset of purpura more commonly had glomerular leukocytes and had the highest percentage of neutrophils among peripheral-blood leukocytes. These findings are consistent with an acute-onset inflammatory reaction in patients with IgAVN who were biopsied <1 month after the onset of purpura. Moreover, serum total IgA concentration correlated with the glomerular mesangial proliferation score. A limitation of this study is the relatively small number of patients from a single center who were of single ethnicity. The patients were followed regularly for up to 6 months after the appearance of purpura at their respective local-area hospitals. However, these data were not been available for this study. This limitation implies that these findings need to be assessed in other cohorts and different ethnic groups, ideally with regular follow-up after the onset of purpura until biopsy as well as after biopsy. Our study demonstrates a chronological association between leukocytic infiltration and glomeruloproliferative responses in IgAVN patients. These findings raise a question whether patients with IgAV should be followed frequently after the onset of purpura to detect nephritis in the early stages and whether assessment of biomarkers, such as Gd-IgA1and Gd-IgA1-specific IgG autoantibodies, should be included.	2021-10-24T15:10:08.499Z	2021-10-22T00:00:00.000	{ "year": 2021, "sha1": "326648ea59e38b86a35ed73fbf2f06076b73bcf6", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2077-0383/10/21/4851/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "8b3be809b4df7128a362b6a48b139e5c7b41ab3c", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
1976278	pes2o/s2orc	v3-fos-license	Osteitis: a retrospective feasibility study comparing single-source dual-energy CT to MRI in selected patients with suspected acute gout Objective Dual-energy computed tomography detects tophi in patients with chronic gout. However, other information that can be obtained from the same scan is not the focus of the current research, e.g., the detection of bone marrow edema (BME) using virtual bone marrow imaging (VBMI). The aim of this study was to evaluate if BME in patients with acute arthritis can be detected with VBMI using magnetic resonance imaging (MRI) as the standard of reference. Materials and methods This retrospective study included 11 patients who underwent both MRI and dual-energy computed tomography (mean interval of 40 days). BME in MRI (standard of reference) and VBMI was judged independently by two different blinded readers. φ-correlation coefficient and Cohen’s κ were performed for statistical analysis. Approval was waived by the IRB. Results Two patients with a final diagnosis of RA and one with septic arthritis showed osteitis on MRI and VBMI. However, in each case, there were individual bones identified with osteitis on MRI but not VBMI. Three additional patients with the final diagnosis of RA were identified correctly as negative for BME. There was a good correlation between both modalities (φ = 0.8; κ = 0.8). Inter-rater reliability was excellent for both modalities (κ = 0.9). Conclusions We have shown that detecting osteitis using VBMI is feasible in patients with inflammatory arthritis. Further studies are needed on larger, more-targeted populations to better define the indications, accuracy, and added value of this technique. Introduction Dual-energy computed tomography (DECT) is known for its ability to detect uric acid tophi in patients with chronic gouty arthritis [1][2][3]. Therefore, it has become a problemsolving tool in patients with unclassified arthritis and suspicion of crystal arthropathy when deemed necessary based on the clinical presentation [4]. However, DECT has many more possible applications that may be useful for clinical practice, including the virtual non-calcium technique that leads to a virtual bone marrow image (VBMI). The latter yields an image that enables detection of bone marrow edema (BME)-previously only possible with magnetic resonance imaging (MRI) [5]. VBMI is based on a three-material decomposition algorithm that allows subtraction of one material (e.g., calcium) to create an image that virtually consists only of two materials (e.g., fat and soft tissue) [5]. Therefore, the resulting images allow differentiation between adult fatty bone marrow with low attenuation in CT (∼ −100 HU) and increased water content (~0 HU) or displacement with the final diagnosis of RA were identified correctly as negative for BME. There was a good correlation between both modalities (φ = 0.8; κ = 0.8). Inter-rater reliability was excellent for both modalities (κ = 0.9). Conclusions We have shown that detecting osteitis using VBMI is feasible in patients with inflammatory arthritis. Further studies are needed on larger, more-targeted populations to better define the indications, accuracy, and added value of this technique. of marrow fat. Using an adjusted soft tissue window, the images display fat as black, while BME appears as gray to white. In the last few years, only a few studies have investigated VBMI, and these have focused on trauma imaging, showing that VBMI is able to characterize the age of vertebral fractures [6,7] and that it is comparable to MRI in displaying bone bruises on the ankle [8]. So far, only one case report published in the literature describes the use of VBMI to detect active inflammation in a patient with suspected sacroiliitis [9]. There is evidence that BME is a strong predictor of progression to bone erosions [10,11] and cartilage damage [12] in patients with rheumatoid arthritis [13,14], even if synovitis is a more sensitive parameter of disease activity. Recent studies were able to show that suppression of osteitis in patients with rheumatoid arthritis goes along with suppression of structural progress, e.g., erosions [15,16]. Studies on bone remodeling suggest that osteitis in MRI is a sign of increased RANKL expression, which leads to osteoclastic bone resorption [17]. However, not all patients with suspected arthritis may be able to undergo MRI of the hands or feet due to various contraindications [18]. Finally, ultrasound has its role in visualizing extraosseous pathology such as synovitis and bone erosion, while intraosseous processes, especially BME, are hidden to the ultrasound probe [19]. We hypothesize that VBMI is able to identify osteitis. Thus, the aim of this study was to evaluate if BME in patients with acute arthritis can be detected with VBMI using magnetic resonance imaging (MRI) as a standard of reference. Materials and methods We retrospectively identified all patients with suspected gout who underwent DECT in the period from September 2011 through July 2014 and additionally had an MRI of the same anatomical region. All patients had an unclear clinical presentation and suspicion of gouty arthritis that occurred either when first presenting at our hospital or later in the course of the treatment. DECT was performed to search for gouty tophi, and MRI in clinical routine to detect inflammatory changes. The report database and hospital PACS (picture archiving and communication system) were used to find those examinations. From the 24 patients who met these criteria, we selected those who had an interval of 3 months or less between DECT and MRI. Eleven patients (six men and five women aged from 45 to 81 years (mean age, 62.2 years) with an interval of 0-82 days (mean, 40 days) between examinations met our criteria and were included. Details of the study population are summarized in Table 1. DECT scans were obtained on a 320-row scanner (Toshiba Aquilion ONE TM and from September 2013 Toshiba Aquilion ONE Vision TM ; Toshiba, Otawara, Japan) with 16-cm z-axis coverage without table movement using 135-kVp (high) and 80-kVp (low) tube voltage [3]. The applied tube current and the resulting radiation exposure depended on the availability of iterative reconstructions (since 2013), patients' physique, and the examined region (see Table 1 for details). For image processing, we used proprietary virtual non-contrast software (dualenergy image view, Version 6, Toshiba, Otawara, Japan) with an adapted dual-energy gradient of 0.69 for calcium. This software is commercially available for Toshiba CT machines, e.g., Aquilion Prime or Aquilion One Vision. Object formulas for the three-material decomposition algorithm were −136/-106 (80/135 kV) for fat and 0/0 for water. A Gaussian noise reduction filter was applied. VBMI images were created in 0.5-mm slices and 5-mm averaged multiplanar reconstructions. We included MR images from different scanners and hospitals (1.5-3 Tesla) acquired using different coils (knee, hand, and flex-coil) into our analysis. An evaluation before study inclusion by reader 1 found all images to be diagnostic and not impaired by severe artifacts, e.g., due to metal implants. The MRI protocol was not the same for all patients but included at least one fat-saturated T2-weighted or fat-saturated proton density (PD) weighted sequence, which can be evaluated for the presence of BME. Slice orientation varied amongst the examinations. However, usually coronal or sagittal images were available. Both readers (reader 1: junior radiologist with 5 years experience in image reading; reader 2: senior radiologist with 14 years experience) independently interpreted the images with an interval of at least 6 months between the reading of MR and VBMI. They used a workstation with a high-resolution monitor and OsiriX Version 5 (Pixmeo SARL, Bernex, Switzerland). The readers were blinded to clinical data, results, and images of the other imaging modality. They had access to all images of the examination including conventional multiplanar CT reconstructions (DECT) or T1 sequences (MRI). Each bone displayed in both imaging techniques was scored as BME-negative or BME-positive. A bone was counted as BME positive if both readers agreed in the presence of edema. Furthermore, the readers documented potential artifacts, e.g., due to incomplete fat saturation in MRI. Statistical analysis was performed using GraphPad Prism (Version 6, La Jolla, CA, USA) and included the calculation of the φ correlation coefficient and Cohen's κ to compare both imaging techniques. Cohen's κ was also used to calculate the inter-rater reliability. The patients' final diagnoses were established by an expert rheumatologist based on clinical presentation, laboratory findings, and imaging results. For patient No. 11, additionally a joint aspiration was performed. The local ethics committee waived the approval of this retrospective study. Results We identified 188 bones that were displayed in both MRI and VBMI. Thirty-eight bones in a total of six patients were excluded because of MRI artifacts with subsequent incomplete fat saturation, resulting in a total of 150 bones that were included in our analysis. All artifacts were caused by inhomogeneity of the magnetic field, e.g., at the tip of the toes. There were no artifacts in the VBMI reconstructions impeding image interpretation. There were five patients with a final diagnosis of rheumatoid arthritis, two with and three without osteitis in MRI, involving ten bones in total. VBMI correctly identified osteitis in both positive patients (seven true-positive, two false-negative, and one false-positive bone) and correctly showed the absence of osteitis in the three negative patients. One patient with infectious arthritis showed osteitis in MRI and VBMI, however, VBMI identified only two out of four affected bones. The bone marrow changes of another patient with bone infarction was detected by both MRI and VBMI. The other patients with the final diagnosis of crystal arthropathy showed no BME in either of the modalities. The false-negative results in VBMI occurred as discreet BME of the proximal phalanx 5 of the tarsus (patient No. 2), the intermediate phalanx 4 (patient No. 3), and the distal fibula and tibia (patient No. 11) in MRI. There was one false-positive detection in the metatarsal head (patient No. 2), as shown in Fig. 1. However, on the patients' level, there were no false-positive or false-negative detections using VBMI. This means, each patient showing BME in MRI was detected by VBMI and each patient without BME in MRI was negative using DECT, respectively. Discussion In our small retrospective feasibility study, we were able to show that VBMI can detect osteitis in arthritis patients. The first results indicate a high sensitivity and specificity compared to MRI as standard of reference. Despite missing 20 % of the edema looking at single bones, all patients with BME were detected by DECT in our small cohort. The depiction of BME using VBMI may be of added value in patients undergoing a clinical DECT scan of the extremities. DECT is an emerging technique for gaining information on material composition and physical properties of a certain tissue. In clinical practice, it is not only used to detect gouty tophi or characterize renal stones [20] but has also evolved into a tool for mapping iodine concentrations and the distribution of contrast media or for virtually subtracting contrast medium or bone of an image [21,22]. DECT analysis of bone marrow using VBMI is another promising application for detecting BME not only in trauma or arthritis but also in malignant bone infiltration, where the technique allows estimating the permeative extension of primary bone tumor cells into normal-appearing bone marrow [23]. VBMI information is obtained by using a three-material decomposition algorithm The φ correlation coefficient was 0.8 and Cohen's κ was 0.8, indicating good agreement of both imaging methods. Inter-rater agreement was excellent for both modalities (MRI κ = 0.87, VBMI κ = 0.88). Figures 1, 2, 3, and 4 provide examples of true-positive, false-positive, and false-negative findings. that is able to virtually eliminate the information of calcium attenuation out of an image. In graphical terms, this can be achieved by projecting all voxels on a graph showing the attenuation of the voxel at high and low kV images parallel to the calcium gradient to a line determined by the specific object formulas of water and fat. The result is an image that is virtually based on the water and fat attenuations only and does not include the calcium information. Therefore, differences in density are related to different water and fat contents allowing to search for BME. The fact that 38 out of 188 bones (or 20 %) in our small collective had to be excluded from analysis due to incomplete fat saturation in MRI indicates that positioning in the MRI machine plays an important role-especially for the appendicular skeleton-and is sometimes complex [24]. Incomplete fat saturation may mimic BME. Therefore, a high signal in fat-saturated MR images must be interpreted with caution. However, VBMI is independent of a homogeneous magnetic field and therefore such artifacts are considerably less frequent. Nonetheless, VBMI uses X-rays and is therefore associated with radiation exposure. However, the investigated appendicular skeleton is far away from the radiosensitive organs such as eye lens or thyroid gland. Furthermore, the examined volume of hands and feet is considerably smaller than for example the abdomen resulting in less radiation for the same image quality. This results in a small conversion coefficient that was recently decreased for the lower extremities using new phantom measurements [25]. In our analysis, we had four false-negative results in VBMI. Mostly, these edemas detected by MRI were rather faint and small in size. Thus, they are more likely to be missed by VBMI. On the other hand, there is a rather long time corresponding fat-saturated T2 sequence -VBMI displays the BME in the fourth metatarsal head (arrowheads) with erosions and cysts in CT. In the first metatarsal head, VBMI is false positive compared to MRI (arrow) Fig. 3 Right wrist of patient No. 7 (CPPD). CT: normal CT image at 135 kV, VBMI: corresponding VBMI image, MRI: corresponding short tau inversion recovery sequence -MRI shows a faint BME in the lunate (arrow) that was missed by reader 1. VBMI also displays a slight hyperdensity, which was scored negative by both readers because even normal carpal bones yield increased density in VBMI compared to long bones according to our clinical experience Fig. 1 Left foot of patient No. 2 (seropositive RA). CT: normal CT image at 135 kV, VBMI: corresponding VBMI image, MRI: corresponding fatsaturated T2 sequencemultiplanar reformatted VBMI shows a good correlation to the corresponding MRI with regard to the identification of BME at the first, second, and fifth metatarsophalangeal joint (arrowheads). However, in MRI, there are artifacts due to incomplete fat saturation in the second and third toes (arrows). The corresponding CT image appearance is normal interval between both imaging modalities in those cases, also in the false-positive detection on the right metatarsal bone. Therefore, the hypothesis can be made that edema might be altered due to treatment effects or worsening of the disease. This study is a feasibility test and suffers from all limitations inherent to a retrospective design including the use of different imaging protocols and variable intervals between the examinations. However, at least for DECT, there is some evidence that radiation dose does not significantly affect the detection of BME [26]. Furthermore, our patient sample is small and heterogeneous, including complex cases and different diagnoses. This is why the first imaging test was inconclusive and both DECT and MRI were performed to secure the diagnosis. This leads to a lack of power, and statistical analysis performed in this study should therefore be treated with caution. Moreover, there was a low overall prevalence of BME in our study. This may be caused by the conservative statistical evaluation counting a bone only as positive if both readers found osteitis. We also did not collect information about the extent and severity of BME in this study. Therefore, it is left to further evaluations to determine if MRI and VBMI are comparable in depicting the degree of osteitis in patients with arthritis. Our study gives first evidence that DECT with VBMI is capable of identifying osteitis in patients with active arthritis, so far only possible with MRI, and thus can provide additional information that may be derived from DECT scans. Furthermore, DECT may be an inexpensive and fast alternative to MRI and a useful supplement to ultrasound. Therefore, prospective data should be acquired to prove the benefit of VBMI in diagnosing arthritis in clinical practice.	2017-08-02T18:47:37.183Z	2016-11-21T00:00:00.000	{ "year": 2016, "sha1": "1c4fb00c28d93725e6f58d395a976d8e0bd0f3a0", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/s00256-016-2533-1.pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "1c4fb00c28d93725e6f58d395a976d8e0bd0f3a0", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
265280759	pes2o/s2orc	v3-fos-license	H3.1K27me1 loss confers Arabidopsis resistance to Geminivirus by sequestering DNA repair proteins onto host genome The H3 methyltransferases ATXR5 and ATXR6 deposit H3.1K27me1 to heterochromatin to prevent genomic instability and transposon re-activation. Here, we report that atxr5 atxr6 mutants display robust resistance to Geminivirus. The viral resistance is correlated with activation of DNA repair pathways, but not with transposon re-activation or heterochromatin amplification. We identify RAD51 and RPA1A as partners of virus-encoded Rep protein. The two DNA repair proteins show increased binding to heterochromatic regions and defense-related genes in atxr5 atxr6 vs wild-type plants. Consequently, the proteins have reduced binding to viral DNA in the mutant, thus hampering viral amplification. Additionally, RAD51 recruitment to the host genome arise via BRCA1, HOP2, and CYCB1;1, and this recruitment is essential for viral resistance in atxr5 atxr6. Thus, Geminiviruses adapt to healthy plants by hijacking DNA repair pathways, whereas the unstable genome, triggered by reduced H3.1K27me1, could retain DNA repairing proteins to suppress viral amplification in atxr5 atxr6. Recent work has revealed that H3.1 interacts with TONSOKU (TSK), which is required for initiating HRR during replication to resolve stalled/broken replication forks and maintain genomic stability 16 .Interestingly, inactivation of TSK in atxr5 atxr6 mutants suppresses heterochromatin amplification, whereas deletions of RAD51 and BRCA1 enhance this phenotype 13,16 .These results suggest that the roles of different HRR proteins in protecting genome stability may be distinct.Furthermore, there is still a gap in our understanding of the interplay between H3.1K27me1 depletion, heterochromatin amplification, and HRR. A growing body of evidence highlights the involvement of HRR factors in viral DNA replication in human [17][18][19][20][21][22] .In plants, the roles of HRR factors in geminiviral propagation are perceived in different ways.Whereas proliferating cell nuclear antigen (PCNA) suppresses the enzyme activity of Rep, RAD54 promotes Rep function in vitro.Moreover, deficiency of PCNA or RAD51D impairs Geminivirus accumulation, but deletion of RAD54 or RAD17 does not affect infection [23][24][25][26][27] .The contrasting reports above indicate that the roles of plant HRR factors in viral DNA replication remain unclear.Of note, when the plant innate immune response is triggered by salicylic acid (SA), RAD51 can directly bind to promoter elements of defense genes and enhance gene expression in a BRCA2-and SA-dependent manner.Moreover, RAD17 and Rad-3-related (ATR) are required to enhance the expression of SA-activated genes and deploy an effective immune response 28,29 .It has been reported that Cabbage Leaf Curl Virus (CaL-CuV) infection can induce the expression of genes in the SA pathway 30 , which in turn seesaws the battle between Geminiviruses and plants 31,32 .These results suggest HRR might regulate the plant defense against Geminiviruses through the innate immune response. We have recently surveyed the viral susceptibility of numerous epigenetic mutants of Arabidopsis.Surprisingly, we found that atxr5 atxr6 double mutants behaved differently and displayed a striking resistance to CaLCuV.Depletion of SAC3B, MBD9, or BRCA1 in atxr5 atxr6 restored susceptibility of the viral infection despite contrasting effects on TE reactivation and heterochromatin amplification.Transcriptome-wide association studies (TWAS) showed that reduced viral DNA replication correlated with upregulation of HRR-related genes in the mutants.We found that the viral protein Rep hijacked host RAD51 and replication protein A 1A (RPA1A) on the viral genome to promote viral amplification.Interestingly, RAD51 and RPA1A showed increased binding to unstable genomic DNA (e.g., rDNA and noncoding RNA (ncRNA) loci) in atxr5 atxr6 vs Col-0.Moreover, RAD51 was enriched at plant defense genes, and its binding was coupled with their transcriptional upregulation in atxr5 atxr6 upon viral inoculation.Additionally, we found that BRCA1, HOP2, and CYCB1;1 recruited RAD51 onto the host genome, and deletion of these factors restored the susceptibility of atxr5 atxr6 to geminiviral infection.Thus, we propose that increased unstable genomic DNA, together with enhanced expression of defense-related gene loci, sequesters RAD51 via BRCA1, HOP2, and CYCB1;1 to prevent the loading of HRR factors onto viral genome, leading to poor viral amplification in atxr5 atxr6.This study provides a new idea to manipulate the routing of plant HRR factors to defend viral infection to improve agricultural traits. As the infection was conducted through agrobacteria-mediated infiltration, we first examined whether the initial plasmid delivery in planta was compromised in atxr5 atxr6.We collected inoculated plants at 3, 6, 9, and 13 days post inoculation (dpi).Southern blot, semi-qPCR, and qPCR results showed that the amount of delivered plasmids in Col-0 and atxr5 atxr6 was comparable at 3, 6, 9, and 13 dpi.In contrast, the amount of replicated viral DNA was strikingly lower in atxr5 atxr6 vs. Col-0 starting at 13 dpi (Supplementary Fig. 2).These results indicated that viral DNA amplification rather than plasmid transfection was suppressed in atxr5 atxr6. ATXR5 and ATXR6 deposit K27me1 specifically on the replicationdependent H3.1 variant, which prevents heterochromatin amplification 10 .H3.1 is encoded by five genes in Arabidopsis thaliana, and inactivation of H3.1 in plants leads to sterility and strong pleiotropic phenotypes 10,16 .Deletion of FASCIATA2 (FAS2), which encodes a subunit of CHROMATIN ASSEMBLY FACTOR 1 (CAF1), prevents the normal deposition of H3.1 during replication 10 .Of note, while fas2 and h3.1 mutants lose both H3.1K27me1 and the H3.1 variant, the atxr5 atxr6 mutants only display a reduced level of H3.1K27me1 without major changes to H3.1 deposition 10 .To examine the impact of H3.1K27me1 deficiency on viral replication, we challenged fas2 and numerous hypomorphic H3.1 mutants with CaLCuV.The fas2 plants showed a more severe yellow mosaic phenotype than Col-0, whereas the H3.1 single mutants and quadruple mutants showed similar infection ratio and symptom severity to those in Col-0 (Supplementary Fig. 3).Of note, inactivation of FAS2 or H3.1 genes did not result in any defect on heterochromatic DNA stability due to the concurrent loss of H3.1K27me1 and H3.1 10,16 .These results imply that concomitant reduction of H3.1K27me1 and H3.1 does not mimic the suppression of viral DNA amplification observed in atxr5 atxr6 where only H3.1K27me1 is reduced.both heterochromatin amplification and TE re-activation in atxr5 atxr6 13,15 .We revisited these experiments using leaves #1-#6 of fiveweek-old mock-treated plants and obtained similar results as the previous report, which used cotyledons in their experiments 13 (Fig. 1c, d and Supplementary Fig. 3g).These triple mutant lines provided an opportunity to investigate the relevance of distinct molecular phenotypes of atxr5 atxr6 with geminiviral pathogenesis in plants. Viral resistance of atxr5 atxr6 is coupled with the enhanced expression of genes involved in DNA repair To pinpoint the genetic pathways that attributed to viral resistance of atxr5 atxr6, we mined public RNA-seq data of mbd9 atxr5 atxr6, sac3b atxr5 atxr6, atxr5 atxr6 and Col-0 from a cotyledon stage to perform transcriptome-wide association studies (TWAS) 13 .Mutations of MBD9 and SAC3B suppress enhanced expression of 240 protein-coding genes in atxr5 atxr6 (Supplementary Fig. 4e, f).Gene ontology (GO) analysis reveals that significant enriched biological processes belonged to immune system process, response to virus and DNA repair (Supplementary Fig. 4g).These three pathways might individually or synergistically contribute to viral resistance of atxr5 atxr6. To investigate how the transcriptome is reprogramed upon virus infection, we performed comprehensive TWAS with high-quality reads (Supplementary Fig. 5, quality score > 30).When the samples from mock and virus inoculation treatments were considered, approximately 4800 differentially expressed genes (DEGs, fold change ≥ 2, FDR ≤ 0.05) were recovered (Supplementary Fig. 6a-c).Among the DEGs, 1136 genes showed enhanced expression upon virus inoculation (Supplementary Fig. 6d, e).Among the virus inoculation-activated genes, we selected 365 genes that were expressed at higher levels in atxr5 atxr6 compared to Col-0, brca1 atxr5 atxr6, mbd9 atxr5 atxr6 and sac3b atxr5 atxr6, in both mock-treated and virus-inoculated samples (Fig. 2a).GO analysis classified the top three enriched biological processes into DNA damage response (DDR,19 genes), DNA repair (17 genes, included in the list of 19 DDR genes) and DNA recombination (12 genes) (Fig. 2b and Supplementary Table 2).Of note, the 18 of 19 DDR genes and 16 of 17 genes related to DNA repair were upregulated in Col-0 upon virus inoculation (Supplementary Fig. 6f).A similar result was also observed in an early microarray assay 30 .Importantly, the expression of DDR genes induced by virus inoculation was further enhanced in atxr5 atxr6; and these DDR genes including HOP2, CYCB1, and BRCA1 belong to HRR rather than NHEJ (Fig. 2c) 34 .Overall, TWAS revealed a significant association between viral resistance of atxr5 atxr6 and enhanced expression of DDR genes. DDR factors are required for efficient amplification of viral genome In our study, virus inoculation activated the expression of 53 genes related to DDR in atxr5 atxr6.Among them, Ataxia-telangiectasia mutated (ATM) and ATR are the kinases that redundantly associate with the majority of DDR factors activated during Geminivirus inoculation (Fig. 2d).One gene, the suppressor of gamma response 1 (SOG1), can govern transcriptional activation of many DDR factors 35 .To decipher the relationship between geminiviral amplification and DDR activation in plants, we performed virus inoculation assays with Col-0, atm, atr, sog1, atxr5 atxr6, and atm atxr5 atxr6 plants.Unlike sog1, atm, and atr showed reduced ratios of symptomatic plants, milder symptoms and less viral DNA compared to Col-0 (Supplementary Fig. 7a-c; Fig. 2e-g).Remarkably, no significant difference was observed in the ratio of symptomatic plants and the amount of viral DNA between atxr5 atxr6 atm and atxr5 atxr6 (Fig. 2f, g).This epistatic phenotype suggests that atm and atxr5 atxr6 function in the same pathway in relation to viral DNA amplification. Rep recruits DNA repair proteins to facilitate viral DNA replication Since ATM is related to HRR, we hypothesized that some HRR factors might directly promote viral DNA amplification.To test this, we conducted yeast two-hybrid (Y2H) screening of 18 selected DDR factors using Rep (Fig. 3a and Supplementary Fig. 7d), the essential viral replication protein, as a bait.Y2H screening recovered RPA1A, RAD51, and PCNA1 as binding partners of Rep (Fig. 3a and Supplementary Fig. 7d).We validated the interaction of Rep with RPA1A and RAD51 through co-immunoprecipitation (Co-IP) experiments in N. benthamiana (Fig. 3b).We hypothesized that if RPA1A and RAD51 were recruited to viral DNA to facilitate viral amplification, they should associate with the viral mini-chromosomes.Indeed, chromatin immunoprecipitation-qPCR (ChIP-qPCR) readily detected the enrichment of RPA1A and RAD51 on the viral genome (Fig. 3c). We also found that the expression of RAD51 and RPA1A was upregulated in atxr5 atxr6 vs. WT and further increased upon virus inoculation in both backgrounds (Fig. 3d).Importantly, mutations in RAD51 or RPA1A resulted in lower ratio of symptomatic plants and reduced viral DNA accumulation compared to Col-0 (Fig. 3e-g).On the other hand, mutations of other HRR factors such as HOP2, BRCA1, or CYCB1 did not affect the viral pathogenesis (Supplementary Fig. 7e, f).These results indicate that RAD51 and RPA1A are essential for Geminivirus amplification in Arabidopsis.Remarkably, loss of RAD51 did not have an additive effect on the viral resistance phenotype of atxr5 atxr6 (Fig. 3f, g).Given that atxr5 atxr6 and atxr5 atxr6 rad51 had the same viral resistance phenotype, we concluded that RAD51 and RPA1A were downstream effectors that accounted for reduced viral amplification in atxr5 atxr6.CaLCuV-inoculation at 9, 11, 13, and 15 dpi.Each dot in the bar plot represents one replicate, the inoculation experiments were performed with 30 plants/replicate with 6 mock-treated plants as a control.Data are presented as mean ± SD (n = 3 biological replicates for Col-0 and atx5 atxr6; n = 1 for atxr5 and atxr6).The experiments were repeated three times with similar results.c Flow cytometry assay shows that loss of MBD9 but not BRCA1 could rescue DNA re-replication phenotype in atxr5 atxr6.d Heat map shows that loss of MBD9, or SAC3B, but not BRCA1, could suppress transcriptional re-activation of TEs in atxr5 atxr6.The quantification was conducted by DESeq2.e The loss of MBD9, SAC3B, or BRCA1 could all increase viral susceptibility of atxr5 atxr6.Photographs of CaLCuV-inoculated plants were taken at 15 dpi.Scale bars, 1 cm.f Percentages of symptomatic plants in different backgrounds induced by CaLCuV-inoculation at 13, 14, and 15 dpi.Each dot in the bar plot represents one replicate, experiments were performed with 36 plants/replicate.Data are presented as mean ± SD (n = 3 biological replicates).g Southern blot assay shows differential accumulation of viral DNA A in CaLCuV-inoculated plants with different genotypes at 15 dpi.Experiments were repeated twice with similar results.The titers of viral DNA A were first normalized with the loading control (EcoRI-digested input DNA, Bottom panel), and then to Col-0 where the amount was arbitrarily set as 1.Statistics in Fig. 1b, f were performed with unpaired twotailed student t-test, , , and *, P < 0.05, 0.01, 0.001, and 0.0001, respectively.Source data are provided in the Source Data File. Over-replicated 45S rDNA in atxr5 atxr6 recruits more RAD51 and RPA1A to the loci A recent study shows that TSK, one upstream HRR factor, preferentially binds to the replication-dependent H3 variant H3.1 to repair the DSBs 16 .Increased DSBs and RAD51 foci have been observed in overreplication associated centers (RACs), a structure observed during remodeling of heterochromatin, suggesting that RAD51 is involved in DNA repair in atxr5 atxr6 12 . We next assessed the distribution of RAD51 and RPA1A signal over the genome in Col-0 and atxr5 atxr6.We first validated the specificity of anti-RAD51 and RPA1A antibodies (Supplementary Fig. 8a-f (ChIP-seq) for the two proteins in Col-0 and atxr5 atxr6.Sample distance clustering analysis of ChIP-seq datasets (Supplementary Fig. 8g) and Venn diagram (Supplementary Fig. 9a, b) showed high reproducibility among replicates, indicating the reliability of our ChIP-seq.When we counted total mapping reads that had multiple mapping locations in genome, we observed numerous peaks in heterochromatin.However, these peaks were not observed when only uniquely mapping reads were aligned (Supplementary Fig. 9c, d).We observed that a majority of RPA1A-bound loci coincided with RAD51-enriched regions, despite the fact that much fewer peaks were identified for the RPA1A ChIP-seq (Fig. 4a, and Supplementary Fig. 9c, d).These results suggested that the two proteins might coordinate with each other during HRR.Indeed, their physical interaction could be readily validated in our Co-IP experiments (Fig. 4b).These results were also consistent with earlier reports showing that concomitant absence of RPA1A and RPA1C mimics rad51 phenotypes (i.e., sterile) 36 .We found that RAD51 and RPA1A were widely distributed over euchromatic regions that contain numerous PCGs (protein-coding genes) and intergenic regions in Col-0 and atxr5 atxr6 (Fig. 4c and Supplementary Fig. 9c-e).Interestingly, this pattern is reminiscent of ChIP-seq patterns of RAD51 and RPA in Mus musculus 37 (Supplementary Fig. 9f), suggestive of their important functions in eukaryotes.Both RAD51 and RPA1A could also bind to heterochromatic regions including rDNA, TEs, and loci corresponding to non-coding RNAs (ncRNAs) (Fig. 4c and Supplementary Fig. 9c-e).We also compared peak numbers in heterochromatic elements in atxr5 atxr6 and Col-0.The overall numbers of RAD51 and RPA1A-bound peaks over rDNA, TEs, and non-coding RNAs (ncRNAs) among the others seemed not to be affected by the loss of ATXR5/6 relative to Col-0 (Supplementary Fig. 9e).One possible reason is that peaks numbers over heterochromatic regions represent a relatively small fraction of the total called peaks.We further performed density profiling of RAD51-and RPA1Aoccupied regions by calculating the fractions of total reads for peaks from the different categories in total reads corresponding to all RAD51and RPA1A-enriched loci.Interestingly, we observed a substantial increase in RAD51 and RPA1A occupancy at the loci corresponding to unknown genes, rDNA (Fig. 4c-f, and Supplementary Fig. 10a), and ncRNAs (Supplementary Fig. 10b, c) in atxr5 atxr6 vs. Col-0.In other words, RAD51 and RPA1A displayed a robust increase in read coverage over rDNA among other classes in atxr5 atxr6 compared to Col-0 (Fig. 4c-f and Supplementary Fig. 10a-c).Of note, the enrichment of RAD51 or RPA1A on TEs was not increased in atxr5 atxr6 vs Col-0, likely because the plant materials were fiveor six-week-old, and the TE amplification does not show an obvious difference in the mutant at this stage (Supplementary Fig. 10d, e). Emerging evidence shows the association between RPA, RAD51, BRCA1, and RAD51-associated protein 1 (RAD51AP1) with transcription processes 38,39 .It has been also shown that reduced H3K27me1 on the 45 S rDNA loci induces the expression of 45 S rRNA variants, which is accompanied by higher copy number of 45 S rDNA in 8 C nuclei of atxr5 atxr6 40 .In our hands, we found that RAD51 and RPA1A were significantly enriched on two sites of rDNA and the enrichment was well correlated with increased copy number of rDNA (Fig. 4d-f, Supplementary Fig. 10d, e) in atxr5 atxr6 compared to Col-0.Furthermore, the increased rDNA and the DNA that transcribes ncRNAs in atxr5 atx6 vs Col-0 were concomitant with reduced level of H3K27me1 (Fig. 4d-f, Supplementary Fig. 10a-e).These results suggested that H3K27me1 might act as a repressive marker to regulate the replication of rDNA, and recruitment of HRR factors onto rDNA.By contrast, reduced H3K27me1 over heterochromatic regions observed in atxr5 atxr6 was restored in mbd9 atxr5 atxr6 and sac3b atxr5 atxr6 mutants 15 .In parallel, the copy number of 45 S rDNA decreased in mbd9 atxr5 atxr6 and sac3b atxr5 atxr6 vs. atxr5 atxr6(Supplementary Fig. 11a).Collectively, these results strongly support a model where reduced H3K27me1 leads to the heterochromatin amplification in atxr5 atxr6, especially rDNA loci, which in turn recruits RAD51 and RPA1A among other HRR factors to repair the increased DNA damage caused by heterochromatin amplification. Coordination between reduced HRR occupancy and impaired transcription in atxr5 atxr6 Besides rDNA loci, RAD51 and RPA1A occupied numerous PCG loci (Fig. 4c-e, and g).GO analysis showed that a large number of RAD51and RPA1A-enriched genes belonged to several genetic pathways, such as response to cold, salt stress, oxidative stress, and jasmonic acid (JA)-mediated signaling (Supplementary Fig. 11b, c).In addition, RAD51 also bound genes related to defense response, response to SA, cell communication, immune system, fatty acid, and other important biological processes (Supplementary Fig. 11b).Interestingly, the PCGs enriched with RAD51 or RPA1A showed very low levels of H3K27me1 in Col-0 and atxr5 atxr6, supporting an active chromatin status at these loci (Fig. 4d, e, g).Further analysis showed that the signals of RAD51 and RPA1A tended to be evenly distributed over gene bodies rather than enriched at promoters in Col-0 and atxr5 atxr6 (Fig. 4d, e, and g).In contrast to rDNA loci, RAD51 signal tended to be reduced over the bodies of PCGs in atxr5 atxr6 vs. Col-0 (Fig. 4d, g). We then selected 366 PCGs that showed lower RAD51 ChIP signal in atxr5 atxr6 vs Col-0 (analyzed by DESeq2, log2 FC < −0.5, P value < 0.05) and assessed their transcription levels.Among the selected genes, the transcripts of 207 PCGs were detectable.Interestingly, 71.0% of the PCGs not only showed decreased RAD51 ChIP signal but also showed reduced transcript accumulation in atxr5 atxr6 relative to Col-0 (Fig. 4h).Thus, our data indicated that reduced occupancy of RAD51 over otherwise actively transcribed regions coincided with their decreased transcript accumulation in atxr5 atxr6.The correlation between DNA repair and transcription observed here is reminiscent of a recent discovery of transcription-associated homologous recombination repair (TA-HRR) 39 .As one detrimental byproduct of transcription, unprocessed R-loops often cause the formation of DSBs, which in turn inhibit local ongoing transcription 41 .Supporting the importance of HRR factors in promoting transcription, RAD51-associated protein 1 (RAD51AP1) induces the formation of R-loop and favors RAD51mediated D-loop formation to restore active transcription over these regions 38 .The fact that normally active chromatin regions in atxr5 atxr6 showed lower RAD51 signals co-occurred with suppressed transcription suggests that the availability of HRR factors is limiting in atxr5 Fig. 2 \| Homology directed repair (HDR) pathway contributes to viral resistance of atxr5 atxr6.a Heat map shows the accumulation change of 365 transcripts, selected from a total of 4800 DEGs based on the clustering analysis, in mocktreated and virus-inoculated Col-0, atxr5 atxr6, sac3b atxr5 atxr6 and mbd9 atxr5 atxr6.The quantification is conducted by DESeq2.b Bubble plots from Gene Ontology (GO) analysis show the enrichment of 365 genes from 2a in different biological processes.c IGV files show changes of transcript levels of indicated genes in mock-treated and virus-inoculated (Ino) Col-0 and atxr5 atxr6 (Normalized by RPKM).Scales for the distinct loci were shown in left as solid lines.Ino, Inoculated.d Protein-protein interaction (PPI) network of DNA repair-related proteins encoded by DEGs upon the virus infection.e Representative phenotypes of CaLCuV-inoculated Col-0, atxr5 atxr6, atm, atm atxr5 atxr6.Photographs were taken at 14 dpi.Scale bars, 1 cm.f Percentages of symptomatic plants induced by CaLCuVinoculation in indicated backgrounds at 12, 13, and 14 dpi.g q-PCR shows the relative amount of viral DNA A in CaLCuV-inoculated plants in indicated backgrounds at 14 dpi.The relative amount of viral DNA A was first normalized to UBQ10 control, and then to that of Col-0 where the mean was arbitrarily assigned a value of 1.In (f) and (g), each dot in the bar plot represents one replicate, experiments were performed with 36 plants/replicate.Data are presented as mean ± SD (n = 3 biological replicates).Statistics In Fig. 2f, g were performed with unpaired two-tailed student t-test, , , * and ***, P < 0.05, 0.01, 0.001, and 0.0001, respectively.Source data are provided in the Source Data File. atxr6 mutants, despite increased expression of these factors in this mutant background. Reduced binding of RAD51 and RPA1A to viral DNA in atxr5 atxr6 The ChIP-seq results suggested that HRR factors are recruited to the unstable genomic elements in atxr5 atxr6.On the other hand, RAD51 and RPA1A among other HRR factors that are de novo viral Rep partners are essential to promote viral DNA amplification.These facts raised the possibility that Geminivirus might compete with host RACs in atxr5 atxr6 for a limiting amount of RAD51, RPA1A and other HRR factors.To test this, we performed RAD51 and RPA1A ChIP-qPCR for the mock-treated and virus-inoculated samples at an early stage.We found that the loading of RAD51 and RPA1A on viral genome, reflected by the ratios of ChIP signal to the viral titer, was significantly higher in Col-0 than that in atxr5 atxr6 (1.69% vs 0.48%) (Fig. 5a).By contrast, we observed a significant enrichment of RAD51 over host rDNA in atxr5 atxr6 than in Col-0, regardless of mock-treated or virus-inoculated samples (Fig. 5a). We next aimed to study how RAD51 and RPA1A were distributed to the host genome.Since atxr5 atxr6 had significantly reduced symptomatic plant rate and resultant lower viral titers (12% amount of Col-0), we purposely increased the number of symptomatic atxr5 atxr6 plants to artificially mimic the symptomatic plant ratio of Col-0 for convenience of ChIP-qPCR assays and ChIP-seq.In this scenario, the virus titers in atxr5 atxr6 could reach to 80% level of Col-0.Interestingly, ChIP-qPCR assays showed that the loading of RAD51 and RPA1A was still significantly decreased over viral genome whereas increased signal of RAD51 was detected over host rDNA and PCG loci in atxr5 atxr6 than that in Col-0 (Fig. 5b).Be noted that the increased loading efficiency of RAD51 and RPA1A onto viral genome vs the natural condition was the trade-off of purposely increasing the number of virusinfected atxr5 atxr6 vs Col-0.Thus, we concluded that RAD51 and RPA1A were indeed poorly recruited to viral genome in the mutant vs Col-0.Using these samples where virus titers in atxr5 atxr6 were purposely increased to 80% level of Col-0, we next performed ChIP-seq of RAD51 and RPA1A to profile their distributions cross host genome in Col-0 and atxr5 atxr6 upon the virus inoculation.We normalized the ChIP-seq reads to the internal control of mitochondrial DNA where RAD51 and RPA1A do not bind so that we could compare the patterning changes of RAD51 and RPA1A associations with host genomes across different samples and treatments. For Col-0, virus inoculation resulted in clearly reduced signal of RAD51 and RPA1A over host genome in virus-inoculated plants vs mock (Fig. 5c, d) despite that the expression of RAD51 and RPA1A was elevated upon virus inoculation (Fig. 3d).The decreased signals of RAD51 mainly originated from the PCG loci and only marginally from rDNA regions (Fig. 5c, d and Supplementary Fig. 11d).This difference was likely due to the lack of DNA re-replication and relative lower DNA damage in Col-0 vs. atxr5 atxr6.The robust loss of RAD51 over PCG loci is also in line with the fact that RAD51 is predominantly associated with PCG loci in a normal condition.Of note, the genes with reduced RAD51 occupancy upon viral inoculation were related to well-known defense pathways involving JA, fatty acid, and s-adenosylmethionine (SAM) (Supplementary Fig. 12a).Moreover, the steady-state transcript levels from those loci were also reduced in inoculated Col-0 plants when compared to the mock (Supplementary Fig. 12b).These results suggested that Geminivirus might suppress transcription of the immune-related genes to attenuate the host defense system, leading to release of RAD51 from the loci and being re-routed onto viral DNA (Supplementary Fig. 12b).Altogether, the accumulated RAD51 in host cells and the RAD51 detained from the genome were all recruited to the viral genome to promote the viral DNA amplification in Col-0 upon virus inoculation (Fig. 3d, Fig. 5c, d and Supplementary Fig. 12a, b). In contrast to Col-0, RAD51 ChIP signal displayed a robust increase in host genome and the signal covered 16865 loci in atxr5 atxr6 upon the virus inoculation (Fig. 5c, d).This result suggested that the accumulated RAD51 was largely recruited onto the host genome in atxr5 atxr6 upon virus inoculation rather than viral genome as observed in Col-0.RPA1A ChIP signal was also significantly distributed onto host genome in atxr5 atxr6 vs Col-0.Despite this, the signal of RPA1A somehow showed a marginally decrease in the genome of atxr5 atxr6 upon the virus inoculation vs mock treatment.This pattern was slightly different from that of RAD51, likely because RPA1A-ChIP-seq only recovered one-tenth of RAD51-bound loci (Fig. 5d).Alternatively, RPA1A has four more orthologs that might surrogate its function. Detailed plotting of ChIP signals showed the increased RAD51 signal covered over rDNA, ncRNAs and PCGs in atxr5 atxr6 upon virus inoculation relative to the mock (Fig. 5d and Supplementary Fig. 12c-e).The rDNA loci harbor relative high copy number of DNA sequences, and are hot spots of DNA damage due to tandem repeat sequences 42 (Supplementary Fig. 10d, e).Thus, over-replicated rDNA and increased DNA damage in atxr5 atxr6 retain RAD51 in the loci 12 .Notably, we found virus inoculation significantly enhanced the RAD51 signal over 270 PCGs (analyzed by DESeq2, Log2 FC > 0.5, P < 0.05) in atxr5 atxr6 whereas only 18.1% of them showed increased RAD51 signal in Col-0 (Supplementary Fig. 13a, Fig. 5a, b).Among 270 PCGs, 177 displayed detectable transcripts.Interestingly, 78% of 177 PCGs also displayed accumulated transcripts in the inoculated atxr5 atxr6 plants (Fig. 5e).These genes are mainly related to defense and immune responses (23 genes), SA (53 genes), and JA (8 genes).Likely, RAD51 contributes to the upregulation of the defense-related loci through a TA-HRR mechanism (Fig. 5e and Supplementary Fig. 13b).Importantly, ChIP signals of RAD51 and RPA1A were significantly enriched over both heterochromatin (exemplified by rDNA) and PCGs in the infected atxr5 atxr6 mutants compared to those of infected Col-0 plants (Fig. 5a-d). Oppositely, the accumulated RAD51 and RPA1A could not be efficiently loaded to the viral mini-chromosomes for amplification in atxr5 atxr6 (Fig. 5a, b).These results suggested that the occupancy of RAD51 and RPA1A in the heterochromatic regions and PCGs both contributed to the viral resistance of atxr5 atxr6 vs Col-0 (Fig. 5a-d).To further test our model, we generated Col-0:35S-FM-RAD51 and atxr5 atxr6:35S-MYC-RAD51 and inoculated the stable transgenic lines with the virus (Supplementary Fig. 13c).Indeed, overexpression of RAD51 in both Col-0 and atxr5 atxr6 clearly increased the ratio of symptomatic plants and viral DNA amount compared to their corresponding controls (Fig. 5f-h).Altogether, we concluded that re-replicated DNA and the loci of defense-related PCGs in atxr5 atxr6 take up HRR components, leading to a shortage of the components essential for viral amplification in atxr5 atxr6 vs Col-0. Increased DNA damage interferes with the interaction between Rep and RAD51 To further validate whether DNA damage attracted more HRR factors onto the host genome and prevented their loading to the virus genome in atxr5 atxr6 vs Col-0, we co-transfected nYFP-HOP2 and cYFP-RAD51 into protoplasts of Col-0 and atxr5 atxr6, then evaluated YFP signal intensity as a proxy of protein-protein interaction level.The complementation of nYFP-HOP2 and cYFP-RAD51 showed YFP signal in the nucleus, reminiscent of the previously reported interaction between HOP2 and RAD51 in mammals 34 .Indeed, a higher YFP signal was detected in the protoplasts of atxr5 atxr6 than those in Col-0, suggesting that more HRR factors are entailed to repair the DNA damage in atxr5 atxr6 vs Col-0 (Fig. 6a, b and Supplementary Fig. 13d).In parallel, we assessed the interaction between RAD51 and Rep in this Fig. 4 \| ChIP-seq assays of RAD51 and RPA1A show that unstable genomic DNA reshapes the distribution of HRR events in atxr5 atxr6 vs. Col-0.a Venn graph shows the overlap between RAD51 and RPA1A enriched regions in host genomes.b Co-IP assay validated the interaction between RPA1A and RAD51 in planta.FM-FVE 77 and Coomassie blue staining of blots serve as negative and loading controls, respectively.The experiments were repeated twice with similar results.c Peak reads distributions of RAD51 and RPA1A ChIP-seq in various locus categories in Col-0 and atxr5 atxr6.The y axis represents the percentage of reads mapped to loci of various categories.ncRNA and TEs represent non-coding RNA and transposable elements, respectively.d, e Distribution of normalized ChIP-signal of RAD51 and RPA1A (normalized with reads of the internal control mitochondrial DNA) and H3K27me1 (RPKM) from Col-0 and atxr5 atxr6 over different categories.H3K27me1 ChIP-seq data were mined from published data (GSE111814).f IGV files of normalized ChIP signals (RPKM) of H3K9me2, H3K27me1, RAD51, RPA1A, H3K27ac, and H3K4me3 on a 45 S rDNA locus on chr2.H3K9me2 and H3K27me1, H3K4me3, and H3K27Ac ChIPseq data were mined from published data GSE111814, GSE166897, and GSE146126, respectively.Scales for the distinct loci are shown on the left as solid lines.g IGV files of normalized ChIP signals (RPKM) of H3K27me1, H3K9me2, RAD51, and RPA1A over selected loci.The bottom panel displayed the IGV files of normalized transcript levels (RPKM) on selected loci from RNA-seq.Scales for the distinct loci are shown on the left as solid lines.h Violin plot shows transcript expression changes from the loci with reduced and unchanged RAD51 ChIP signal in mock-treated atxr5 atxr6 vs Col-0.Horizontal lines in the bar plots display the 75 th , 50 th and 25th percentiles, respectively.Whiskers represent the minimum and maximum values.P value is calculated by unpaired two-tailed Welch's approximate t-test.Source data are provided in the Source Data File.13e).These results, together with our ChIP-seq, indicated that HRR factors would be efficiently sequestered to the genome of atxr5 atxr6, preventing them from being recruited to viral genome for amplification. We next assessed whether the scenario that DNA damage could retain HRR factors in host DNA, preventing the viral replication could go beyond atxr5 atxr6 mutants.It has recently been reported that transgenic plants expressing H3.1S28A, but not H3.1S28A31T, show heterochromatin amplification and activation of DNA repair pathway 16 .In line with this, we could readily detect significantly increased rDNA in 6e).Importantly, the H3.1S28A transgenic plants showed robust virus resistance against CaLCuV, whereas the H3.1S28A A31T lines did not (Fig. 6f-h).Thus, in the H3.1S28A line, unstable genomic elements could also retain HRR factors and prevent them from being recruited onto the viral genome, leading to suppression of viral DNA amplification.Since SA triggers DNA damage and also induces the expression of defense genes, we consider this treatment as a natural condition, reminiscent of the physiological condition in atxr5 atxr6.In our hands, the exogenous SA treatment activated the DNA damage response and suppressed the viral DNA amplification (Fig. 6i, j).Moreover, the application of DNA damage reagent bleomycin (BLM) enhanced the immunity of wild-type plants against virus inoculation (Supplementary Fig. 14a, b).These results consolidated our model that DNA damage induced by unstable DNA regions can request HRR factors to the host genome to prevent them from being used by the viral genome for efficient viral DNA amplification. BRCA1, HOP2 and CYCB1 are bona fide partners of RAD51 We next aimed to identify the factors that potentially recruited RAD51 or RPA1A to the host genome.Through Y2H assays, we recovered BRCA1, and CYCB1;1 as partners of RAD51 from a few candidate genes (Fig. 7a, and Supplementary Fig. 14c).Neither protein appeared to directly interact with RPA1A in the assays (Fig. 7a, b).RAD51 could be readily detected in the co-immunoprecipitated products of CYCB1;1 and BRCA1 but not in control IPs (Fig. 7b).In addition, homologous pairing protein 2 (HOP2) interacted with RAD51 in Co-IP, but not in Y2H assay (Fig. 7b, Supplementary Fig. 14c).Of note, all three proteins are intrinsically disordered or contain disordered segments; and these features might contribute to their interaction with RAD51 (Supplementary Fig. 14d). Roles of BRCA1, HOP2 and CYCB1 in loading RAD51 Interactions of BRCA1, HOP2, and CYCB1;1 with RAD51 detected here and also in mammalian cells and plants 34,43,44 suggested the potential roles of these proteins in the recruitment of RAD51 onto unstable DNA.A robust increase of viral DNA was detected in brca1 atxr5 atxr6 compared to that in atxr5 atxr6, and no obvious difference in the viral titer was detected between brca1 and Col-0 (Fig. 1e-g).These contrasting results suggested that RAD51-centered DNA repair components were not able to reach the unstable loci in atxr5 atxr6; rather, the components were re-routed to viral mini-chromosomes for DNA amplification in brca1 atxr5 atxr6 compared to atxr5 atxr6 (Fig. 7c, d).Differently, DNA repair components were largely recruited to viral genome in the single mutant brca1 and Col-0 where host DNA is intact in a normal physiological condition.Thus, BRCA1 might regulate viral DNA amplification in a manner dependent on RAD51 and other HRR factors in the atxr5 atxr6 background. To test whether deletions of HOP2 and CYCB1;1 would also interrupt recruitment of HRR factors between host DNA and viral genome in atxr5 atxr6, we generated cycb1 atxr5 atxr6 and hop2 atxr5 atxr6 mutants through genetic crossing and tested their susceptibility to viral inoculation.When challenging the higher order mutants with CaLCuV, hop2 atxr5 atxr6 and cycb1 atxr5 atxr6 mutants, like brca1 atxr5 atxr6, had higher percentages of symptomatic plants and more severe chlorosis phenotype than those in atxr5 atxr6 (Fig. 7e-g).Moreover, hop2 atxr5 atxr6 and cycb1 atxr5 atxr6 mutants showed higher viral DNA content relative to atxr5 atxr6 (Fig. 7e-g).These results indicated that BRCA1, HOP2, and CYCB1;1 were essential for the virus resistance phenotype in atxr5 atxr6, with the suggestion that the three proteins might be responsible for the recruitment of RAD51 to the unstable loci and PCGs in the host genome.Of note, differential susceptibilities among the triple mutants (Fig. 7e-g) suggested that their contributions to the loading of RAD51 or other HRR factors onto damaged DNA might be different.Alternatively, additional paths for recruitment to RAD51 and other HRR factors might compensate for the loss of the three components in different degrees. To further test our model, we selected several representative loci on heterochromatic regions and PCG loci to perform ChIP-qPCR.Indeed, RAD51 signals over heterochromatic regions including 5.8 S rDNA were all lower in the triple mutants than that in atxr5 atxr6 under mock treatment and inoculation conditions.This result indicated the deficient recruitment of RAD51 onto host heterochromatic regions in the triple mutants vs atxr5 atxr6 (Fig. 7c and Supplementary Fig. 14e).Moreover, deletions of CYCB1;1, HOP2 and BRCA1 in atxr5 atxr6 also prevented the loading of RAD51 on PCGs upon virus inoculation (Fig. 7c).On the other hand, more efficient loading of RAD51 on viral genome was observed in hop2 atxr5 atxr6, brca1 atxr5 atxr6 and cycb1 atxr5 atxr6 compared with that of atxr5 atxr6 (Fig. 7d).In contrast, there was barely heterochromatin amplification and HRR events in WT background.As a consequence, the amount of RAD51 was sufficient for viral amplification, deletions of CYCB1;1, HOP2 and BRCA1 did not affect the viral pathogenesis and accumulation (Supplementary Fig. 7e, f).Thus, we concluded that BRCA1, HOP2 and CYCB1;1 indirectly regulate viral DNA replication through controlling the availability of RAD51 to the viral genome in the atxr5 atxr6 background (Fig. 7h). Discussion Here we reported a novel regulatory role of histone MTases on DNA virus amplification.Different from other histone MTases that deposit repressive marks on viral genomes and typically repress viral infections, ATXR5 and ATXR6 coincidentally promote viral amplification as a trade-off to maintaining host genome integrity.This unique mode of action in virus-inoculation is highlighted by the fact that loss-offunction mutants of atxr5 atxr6 become resistant to the virus compared to Col-0.The underlying mechanism for the viral resistance is that mutations of ATXR5 and ATXR6 cause unstable heterochromatic regions exemplified by rDNA and ncRNAs loci, which activate the HRR pathway.These unstable heterochromatic regions together with defense-related PCGs in euchromatin regions, in turn, sequester HRR P value is calculated by unpaired two-tailed Welch's approximate t-test.f Representative pictures of the virus-inoculated plants in indicated lines.Photographs were taken at 14 dpi.Scale bars, 1 cm.g Percentages of symptomatic plants of virus-inoculated plants in indicated backgrounds at 11 and 13 dpi.h Constitutive expression of RAD51 promotes the viral DNA amplification in Col-0 and atxr5 atxr6.qPCR assays show increase of virus titers in RAD51 overexpression lines vs their reference backgrounds.Samples were collected at 14 dpi.Normalization of viral DNA was conducted as Fig. 2g.In (a), (b), (g) and (h), each dot in the bar plot represents one replicate, the experiments were performed with 36 plants/replicate.Data are presented as mean ± SD (n = 3 biological replicates).Statistics in Figs.5a, b, g, and h were performed with unpaired two-tailed student t-test, , , * and ***, P < 0.05, 0.01, 0.001, and 0.0001, respectively.Source data are provided in the Source Data File.factors and preclude them from being hijacked by the virus-encoded Rep protein, leading to suppression of the viral replication (Fig. 7h).Several pieces of evidence support our model: 1) DNA repair factors such as ATM is required for efficient viral DNA amplification (Fig. 2e-g loading of RAD51 and RPA1A on viral genome is detected in atxr5 atxr6 relative to Col-0, despite higher accumulation of the proteins in the mutant and stronger ChIP signal over the atxr5 atxr6 genome relative to Col-0 (Fig. 5a-d); and 6) BRCA1, HOP2, and CYCB1;1 promote the recruitment of RAD51 onto the host genome, whereas depletions of HOP1, BRCA1, and CYCB1 in atxr5 atxr6 cause inefficient loading of RAD51 onto the unstable host genome and PCGs related to defense, and re-routing of RAD51 onto viral genome in atxr5 atxr6, leading to hyper-susceptibility to virus inoculation in the triple vs atxr5 atxr6 (Fig. 7c-g).Thus, we conclude that RAD51 and RPA1A, among other HRR factors, are the cornerstones in the battle of host and virus. If the virus fails to recruit RAD51 and RPA1A, it will fail to parasitize the host, as seen in atxr5 atxr6. DSBs activate HRR to faithfully repair DNA during replication to maintain genome stability 45 .As parts of heterochromatic elements, rDNA is considered a hot spot for recombination events due to their repetitive elements.During HRR, the RPA complex coats resected ends to prevent their degradation.RAD51 then replaces RPA and mediates the D-loop formation and strand invasion 45 .Robust loading of RAD51 and RPA1A onto rDNA and ncRNAs loci accompanies lower H3.1K27me1 levels in atxr5 atxr6 (Fig. 4d, e).Similarly, reduced H3.1K27me1 correlates with higher rDNA and ncRNA copy number (Fig. 4d, e, Supplementary Fig. 10b-e).These results suggest that RAD51 and RPA1A participate in maintaining genome stability at heterochromatic regions in the absence of H3.1K27me1.In our study, BRCA1 and HOP2 promote the recruitment of RAD51 over rDNA regions to facilitate DNA repair in atxr5 atxr6.A similar mechanism has been reported in mammalian cells, where BRCA1 and HOP2 complexes stabilize the RAD51-ssDNA filament and promote RAD51-mediated homologous DNA pairing process in vitro 34,44 .In Arabidopsis, CYCB1;1-CDKB1 complex phosphorylates RAD51 in vitro 43 .Phosphorylation of RAD51 is required for its efficient DNA binding 46 .Remarkably, loss-offunction mutations of BRCA1, HOP2, and CYCB1;1 in atxr5 atxr6 compromised viral resistance of atxr5 atxr6 to different extents (Fig. 7c-g), suggesting robust loading of RAD51 over heterochromatic region in atxr5 atxr6 mutants is required for its viral resistance. RAD51 is also distributed along the gene body of PCGs with low H3K27me1 under physiological conditions (Fig. 4d).Furthermore, decreased RAD51 ChIP signals over PCGs are concordant with reduced transcripts levels in virus-inoculated Col-0 (Supplementary Fig. 12b).Notably, the RAD51-enriched PCGs are dedicated to the defense response, response to hypoxia, SA, JA and cold and salt stresses.These observations suggest that RAD51 might coordinate TA-HRR to regulate the transcription, while coping with physiological stresses (Supplementary Fig. 11b).Exogenous SA treatment induces DNA breaks and promotes RAD51 binding to the promoters of pathogenesis-related genes (PR).Conversely, depletion of BRCA2A 28 and ATR 29 , two partners of RAD51, suppresses the expression of SA-induced defense-related genes and efficient immune response.These results further imply that RAD51 is required for efficient transcription regulation and defense response.Importantly, the fact that PCGs display reduced occupancy of RAD51 while producing lower expression of defense-related transcripts in infected Col-0 plants implies a dual role of Rep during infection (Supplementary Fig. 12a, b): hijacking the HRR factors to facilitate the viral genome amplification whereas attenuating the plant defense system in WT plants.By contrast, RAD51 occupancy over defense-related genes and transcription of the corresponding loci were enhanced in atxr5 atxr6 upon virus inoculation (Fig. 5c-e, Supplementary Fig. 13b), implying that the elevated transcription of defense genes in atxr5 atxr6 compete with the virus for a limited amount of RAD51, restricting virus amplification and enhancing the immune response, thus resulting in viral resistance.During the parasitic lifestyle, DNA viruses activate the host DNA repair mechanism and hijack the replication machinery to the viral genome.Here, rad51 and rpa1a mutants displayed resistance to Geminivirus inoculation, indicating that the proteins are critical for viral amplification.Consistently, RAD51 interacts with the Rep encoded by mungbean yellow mosaic India virus (MYMIV) and promotes geminiviral DNA replication in the heterologous system Saccharomyces cerevisiae 47 .Geminiviruses replicate their genomes through rolling circle replication and recombination-dependent replication 2 .RAD51 might function as a recombinase to directly facilitate the rolling circle replication of the virus 48 .RPA1A might also contribute to the rolling circle replication of geminiviruses, reminiscent of RPA-mediated enhancement of the replication of simian virus 40 49,50 .As key components for homologous recombination, RAD51, and RPA1A might also directly promote the recombination-dependent replication of Geminiviruses.Supporting this model is that RAD51D, a paralog of RAD51, has been reported to promote the recombination-dependent replication of Geminivirus 51 .In our studies, although virus inoculation enhances the expression of RAD51 and RPA1A in Arabidopsis, the ChIP signals of RAD51 and RPA1A over the host genome were reduced in inoculated vs mock-treated Col-0 plants (Figs.3d and 5c, d).Furthermore, Rep interacts with RAD51 and RPA1A in planta (Fig. 3b).All these observations clearly indicate that Rep hijacks RAD51 and RPA1A to facilitate viral amplification during the infection process. In summary, we propose that robust retention of RAD51 and RPA1A onto unstable host DNA, along with increased RAD51 accumulation over the upregulated defense-related PCGs, prevent the efficient loading of the replication-essential factors onto the viral genome, leading to a resistance phenotype in atxr5 atxr6.This study implies that the virus might adapt to the healthy host to hijack host DNA-repairing components for the viral replication.On the other hand, an unstable genome could retain DNA-repairing proteins onto host genome to gain plant resistance to viral infection.One implication of this study is that one might apply a certain level of genome toxicity stress on crops or insert certain pieces of RAD51-favored unstable DNA elements into crop genomes 52 .Such actions might be able to trap HRR factors onto host genomes to a certain degree, while granting viral resistance, ensuring agricultural production. Fig. 6 \| Increased DNA damage promotes RAD51 interaction with host HOP2 but not viral Rep. a BiFC of YFP assays showed that mutations of ATXR5/6 promote the interaction between RAD51 and HOP2.See Supplementary Fig. 13d for negative controls.15 independent protoplasts were examined for the interaction and showed similar results.Scale bars, 5 μm.b Statistical analysis of YFP signal intensity of each protoplast.c BiFC of YFP assays showed that interaction between RAD51 and Rep was compromised in atxr5 atxr6.See Supplementary Fig. 13e for negative control.15 independent protoplasts were examined for the interaction and showed similar results.Scale bars, 5 μm.d Statistical analysis of YFP signal intensity of each protoplast.In (b) and (d), each dot in the bar plot represents one protoplast.Data are presented as mean ± SD (n = 15 biologically independent prototplasts.e q-PCR assays show the amount of rDNA in different lines.The relative amount of rDNA was normalized against UBQ10.f Representative phenotypes of CaLCuV-inoculated Col-0, atxr5 atxr6, and third generation of (T3) lines expressing H3.1 variants.Photographs were taken at 14 dpi.Scale bars, 1 cm.g Percentages of symptomatic plants induced by CaLCuV-inoculation in indicated backgrounds at 10, 12, and 14 dpi.h q-PCR assays show the amount of viral DNA A in CaLCuVinoculated plants at 14 dpi.Normalization of viral DNA was conducted as Fig. 2g.i SA treatment activates the DNA damage response.qRT-PCR assays show the transcription level of HRR factors in indicated genotypes and treatments.j q-PCR assays show the relative amount of viral DNA A in the indicated genotypes and treatments at 14 dpi.In (e), (g), (h), (i), (j), each dot in the bar plot represents one replicate.Experiments were performed with 36 plants/replicate except in (i) 6 plants/replicate was used.Data are presented as mean ± SD (n = 3 biological replicates).Statistics In Figs.6b, d, e, g, h, i and j were performed with unpaired twotailed student t-test, , , * and ***, P < 0.05, 0.01, 0.001, and 0.0001, respectively.Source data are provided in Source Data File. Similarly, disabling H4K20me1 transferases in human, the functional homologs of ATXR5/6, can also activate TONSOKU-LIKE and BRCA1 machinery 53,54 .Host retention of the DNA repairing machinery to correct genome instability would be expected to increase host immunity and impair the DNA virus amplification, thus providing an opportunity to defend against the viruses.In addition, breaks on rDNA lead to the loss of rDNA repeats and exacerbate genomic instability during aging 55,56 .Overexpression of RAD51 restores the accumulated DSBs in aged cells and extends the replicative life span of yeast 56,57 , implying potential application of H4K20 and RAD51 in curing rDNA-associated human diseases such as amyotrophic lateral sclerosis, and Huntington's disease 58 .confirmed by enzyme digestion were transferred to GV3101 for transient expression in N. benthamiana. For Bimolecular fluorescence complementation (BiFC) constructs, full-length CDSs of Rep and HOP2 were cloned into pBA-nYFP-DC by LR reaction.RAD51 CDS was cloned into pBA-cYFP-DC.Plasmids were confirmed by enzyme digestion, extracted as previously described 14 , and transfected to protoplasts. For stable transgenic plants, pBA-Flag-4Myc-RAD51 was obtained as described above.To obtain pCambia1300-Myc-RAD51, pCambia1300-Myc-DC-Nluc was digested by PstI and dephosphorylated with Calf Intestinal Alkaline Phosphatase (CIP).The missed destination cassette DC fragments was obtained by PCR of another normal DC cassette followed by PstI-digesttion.The resultant fragment was ligated into PstI/CIP treated pCambia1300-Myc-DC-Nluc to generate pCambia1300-Myc-DC.Full length CDS of RAD51 was transferred into pCambia1300-Myc-DC by LR reaction. CaLCuV inoculation assays Plants were grown on MS plate under 8-h light/16-h dark conditions for 10 days and then transferred to soil under 8-h light/16-h dark conditions for around 14 days to reach eight true-leaf developmental stage.We first extracted the plasmid from E. coli (DH5α) that contains the geminivirus genome and transformed it into agrobacteria.Fresh agrobacteria containing geminivirus genome together with silicon carbide powder was sprayed by a pump onto the center of Arabidopsis (which were newly emerged rosette leaves) at 80 psi (pound per square inch).Plants with different genotypes were infected by agroinfiltration of CaLCuV infective clones of pNSB1090 DNAA and pNSB1091 DNAB.The symptoms of plants were monitored and evaluated daily since we began to observe the yellow mosaic and chlorosis in viral infected plants.The SA treatment was conducted as previously described 60 .The concentration of bleomycin (BLM) treatment was selected as previously described 29 and carefully conducted in a hood with proper protection.To assess systemic infection, we harvested the eight newly emerged rosette leaves of CaLCuV-inoculated plants.We collected 36 CaLCuV-inoculated plants as one biological replicate to perform the following assays for the majority results.For rad51 and rad51 atxr5 atxr6, we collected 24 or 15 homozygous plants as one biological replicate (rad51 is sterile, genotype is confirmed by PCR after viral inoculation) respectively.We collected 36 CaLCuVinoculated plants and Mock-treated Col-0 and atxr5 atxr6 as one biological replicate to perform the assays in Fig. 6a.For the assays in Fig. 6b, we collected 36 Mock-treated Col-0 and atxr5 atxr6 and 36 CaLCuV-inoculated Col-0 as one biological replicate.To purposely increase the symptomatic plants in CaLCuV-inoculated atxr5 atxr6, we first inoculated 360 atxr5 atxr6 together with 108 Col-0 (36 Col-0 in one tray as one biological replicate), and selectively collected atxr5 atxr6 with CaLCuV-induced phenotypes from 360 CaLCuV-inoculated atxr5 atxr6 to make sure the number of symptomatic plants in the atxr5 atxr6 (selected 36 CaLCuV-inoculated plants as one biological experiments were performed with 32 plants/replicate.Data are presented as mean ± SD (n = 3 biological replicates).h A proposed model for Geminivirus competition with host genome for HRR to facilitate viral amplification.In the WT plants with a stable genome the virus inoculation suppresses the transcription of defenserelated genes to evict RAD51 and RPA1A from host genome, and then viral-encoded Rep protein hijacks RAD51 and RPA1A for efficient viral replication.However, in atxr5 atxr6, unstable host DNA retains large amount of DNA repairing factors and promotes the transcription of defense-related genes.Consequently, unstable host heterochromatic elements coordinate with the upregulated defense-related genes to retain a large amount of DNA repairing machinery to prevent its rerouting to viral genome, leading to low-efficiency virus amplification.Figure 7h was created with BioRender.com.Statistics in Figs.7c, d, f, and g were performed with unpaired twotailed student t-test, , , * and ***, P < 0.05, 0.01, 0.001, and 0.0001, respectively.Source data are provided in the Source Data File. replicate) was the same to that in Col-0.To assess the plasmid transfection efficiency, we collected the 12 CaLCuV-inoculated whole plants except for cotyledons at 3, 6, 9, 13, and 16 dpi. Flow cytometry Flow cytometry profiles were generated as described 14 with some modifications.We collected around 0.3 g rosette leaves from mocktreated plants with distinct genotypes at 14 dpi and finely chopped in 3 ml freshly made nuclear extraction buffer (45 mM MgCl 2 , 30 mM sodium citrate, 20 mM pH 7.0 MOPS, 0.1% Triton X-100, 5 mM sodium metabisulfite, 5 μl/ml β-mercaptoethanol, 100 μg/ml RNaseA) and filtered through 40 μm cell strainer (Sigma) to release the nuclei.Nuclei were stained by adding 150 μl 1 mg/ml propidium iodide (Sigma) with gentle mixing by pipetting 10 times.Flow cytometry profiles were obtained from BD FORTESSA X-20 (College of Medicine Cell Analysis Facility, Texas A&M). Western blot analyses Western blot analyses were performed as previously described 61 .Membranes were first incubated with antibodies against Myc (Sigma, C3956), H3 (Agrisera, AS10 710), or HA (Sigma, H9658) and then incubated with goat-developed anti-rabbit (GE Healthcare, NA934) or goat-developed anti-mouse immunoglobulin G (GE Healthcare, NA931).Membranes were developed with ECL+ and signals were obtained using ChemiDoc XRS+ and analyzed by Image Lab software (Bio-Rad). Strand specific RNA sequencing library preparation Total RNA was extracted from virus-inoculated and mock-treated plants at 16 dpi with TRIzol and then treated with TURBO Dnase (Thermo Fisher, AM2238).Messenger RNA was enriched with Dyna-beads™ Oligo(dT) 25 (Invitrogen TM , 61005) according to the manufacturer's manual.First strand synthesis was completed with Random primer (Invitrogen™, 48190011), RNaseIn (Invitrogen™, AM2696), SuperScript™ III Reverse Transcriptase (Invitrogen™, 18080044) after fragmentation.Second strand synthesis was conducted with dUTP mixture (20 mM dUTP, 10 mM dATP, dCTP, and dGTP), RNase H (NEB, M0297), DNA polymerase I (NEB, M0209) and E. coli DNA ligase (NEB, M0205).End repair was conducted with NEBNext Ultra TM End repair/ dA-Tailling Module (NEB, E7442) followed by the adapter ligation with ligation mix and adapter (Illumina, TruSeq Kits, 15026773) according to the manufacturer's manual.Enriched mRNA from 1 μg total RNA was used as the starting material and 15 cycles were used to amplify the library. Anti-RAD51 or anti-RPA1A were conjugated to protein A magnetic beads (Thermo Fisher, 88845) with BS3 crosslinker (C 16 H 18 N 2 O 14 S 2 Na 2 ) as previously described with minor modifications 62 .Protein A magnetic beads (150 μl) were washed with nuclei dilution buffer five times, incubated with 15 μg anti-RAD51 or anti-RPA1A antibodies in nuclei dilution buffer overnight at 4 °C, and then washed with nuclei dilution buffer.The antibodies were conjugated to the beads with freshly prepared crosslink solution with 5 mM BS3 at room temperature for 1 hr.The reaction was terminated with 60 mM Tris-HCl pH 7.5 at room temperature for 20 mins.Then the beads were washed with nuclei dilution buffer three times and used for immunoprecipitation with isolated nuclei (4 °C for 4 hours).Finally, the immunoprecipitates were washed with nuclear dilution buffer three times and eluted with elution buffer (5 mM EDTA, 200 mM NH 4 OH) for western blot analysis. BiFC assay The transient gene expression in protoplasts from Col-0 and atxr5 atxr6 for BiFC assays were performed as previously described 63 .pBA-cYFP-RAD51 was co-expressed with pBA-nYFP-Rep and pBA-nYFP-HOP2 in the protoplasts.Sixteen hours after transfection, the fluorescence signals were captured and evaluated with Leica SP8 confocal microscope.The excitation light wavelength for YFP and chlorophyll autofluorescence was 514 nm and 633 nm, respectively.At least 15 individual protoplasts were examined for each transformation to obtain similar results. Quantitative PCR and qRT-PCR The relative amount of viral DNA A in viral-infected plants and endogenous rDNA of Arabidopsis were examined by q-PCR.High-quality DNA was obtained as described in southern blot analyses.Ubiquitin 10 served as an internal control for normalization.The enrichment levels of RAD51 and RPA1A on specific loci on viral genome and Arabidopsis after ChIP assays were also assessed by q-PCR.The q-PCR assays were performed with 10 μl system containing 5 μl SYBR Green Master Mix in 384-well plate.PCR conditions, signal detection, and quantification were performed as previously described 4 . Total RNA was extracted with TRI reagent (Sigma) from adult plants with eight true leaves.Normalized RNA was treated with Turbo DNases and then reverse-transcribed with reverse transcriptase (SuperScript III, Invitrogen).Reverse transcription was primed by oligo(dT).The following q-PCR assays were performed with 10 μl system containing 5 μl SYBR Green Master Mix in 384-well plate.PCR conditions, signal detection, and quantification were performed as previously described 4 . ChIP sequencing library preparation Half of ChIP-enriched DNA or one sixth of input DNA were used as starting materials for DNA end repair process for library construction with NEBNext Ultra TM End repair/dA-Tailling Module (NEB, E7442).Then the products were ligated to adapters (NEBNEXT Multiplex Oligos for Illumina, E7335, E7500 and E7710) with Blunt/TA Ligase Master Mix (NEB, M0367) according to the manufacturer's manual.For the library preparation, 9 cycles were used to amplify products from input DNA, 13 cycles were used to amplify RAD51 ChIP-enriched DNA and 16 cycles were used to amplify RPA1A ChIP-enriched DNA. Illumina sequencing and analysis Libraries for strand-specific RNA sequencing were sequenced on the Novaseq system with pair-end 150 bp read length (Novogene).Adapters trimming, mapping, and counting process were performed as previously described 14 .Normalization and differential expressed analysis were calculated by DESeq2 (ver.3.15) 64.The visualization for selected loci was shown in Intergrative Genomics Viewer (IGV) 65 . Heat-map clustering was performed based on sample-to-sample distances.To obtain the comprehensive DEG lists which contain both mock-treated and viral infected samples, we first compared mock-treated atxr5 atxr6 to mock-treated Col-0 using Col-0 as a reference and then compared the other three triple mutants to atxr5 atxr6 using atxr5 atxr6 as a reference.The same analysis was also performed on viral-infected samples.Moreover, we also compared the viral infected Col-0 to mock-treated Col-0 using as a reference, then compared the other viral infected genotypes to their mock-treated samples.The DEGs were selected using the cut-off (Fold change > 2 and p < 0.05), but the genes with low expression (max reads > 10) were filtered.The pool of DEGs from the three groups mentioned above were the comprehensive DEG lists to perform heatmap clustering analysis. Libraries for ChIP-seq were sequenced on the Novaseq system with pair-end 150 bp read length (Novogene).Adapter trimming and mapping processes were performed as previously described 14 .The leftover reads later were mapped by bowtie2 (ver.2.4.4) 66 with perfect matches using Arabidopsis genome TAIR10 (http://www.arabidopsis.org/) as a reference genome.Reads uniquely mapped to the genome and reads mapped to multiple locations in the genome were separately extracted by Samtools (ver.1.15.1) 67for distinct downstream peak calling analyses.MACS2 (ver.2.2.5) 68 was used for ChIP-seq peak calling using default parameters, both narrowPeak and broadPeak were called for each sample.We used the reads mapping to the genome which contains sequence mapped to multiple locations in the genome to perform the ChIP signal profile, peak number, peak reads percentage analysis and comparative analysis with DESeq2 (ver.3.15) 64since the analysis with uniquely mapped reads barely detected signals in the majority over-replicated regions that harbor many repetitive sequences in atxr5 atxr6 9 . Peaks with P-value < 0.05 were selected for the following analysis, narrowPeak and broadPeak were merged by bedtools 69 (ver.2.29.2).Annotation of peaks was conducted using bedtools.with TAIR10 gtf file as a reference.Peaks from biological replicates for the same genotype were merged and classified into different categories to calculate the relative peak number percentage. The peaks belonging to distinct categories were first extracted with Samtools 67 , and then reads over peaks from distinct categories were counted with Subread featureCounts (ver.2.0.0) 70.Total reads of peaks from specific categories were divided by total reads of peaks from all categories to generate peak reads percentages for individual specific categories.Normalization of the ChIP-seq signals was performed with concepts adopted from THOR 71 .Here, mitochondrial DNA was selected as "housekeeping" controls for the normalization of reads.Be noted that the reads mapping to mitochondria were relatively stabilized from 1.92% to 2.11% of total reads among our input samples and thus could serve as internal "housekeeping" controls.Furthermore, none of 16865 RAD51-enriched or 1675 RPA1A-enriched peaks were mapped into the mitochondrial genome, indicative of successful enrichment of RAD51 and RPA1Abound chromatin in nuclei).The normalization based on mitochondrial (mt) DNA was performed first by counting mitochondrial reads in each sample.The percentage of the mtDNA reads in total mapped reads were listed in Supplementary Table 3.The scaling factor for each sample was calculated as the ratio of mtDNA reads /that of maximum mtDNA reads among all IP samples.Scaling factors were then applied to perform normalization by multiplying with total mapped read counts.Be noted that the ratios of mtDNA input in all samples were essentially the same and thus not used for the normalization.All downstream statistical analyses and plotted graphs were generated by R (ver.4.0.2) and ggplot2 72 .All ChIP-seq profiles for RAD51, RPA1A, and H3K27me1 were drawn by deeptools (ver.3.7.4) 73using the default parameters.DESeq2 (ver.3.15) 64with default parameters (P-value < 0.05) was implemented to output differential peaks between different groups of ChIP-seq samples.All violin plots were plotted by ggviolin package.All the plots and profiles for ChIP-seq were drawn by R. The visualization for selected loci was shown in Intergrative Genomics Viewer (IGV) 65 . Fig. 1 \| Fig. 1 \| atxr5 atxr6 displays robust viral resistance phenotype that is superficially uncoupled with re-activation of TEs and DNA replication.a Loss-offunction mutants of atxr5 atxr6 show resistance to CaLCuV.Photographs of CaLCuV-inoculated Col-0, atxr5, atxr6, and atxr5 atxr6 plants at 16 dpi (days postinoculation).Scale bars, 1 cm.b Percentages of symptomatic plants induced byCaLCuV-inoculation at 9, 11, 13, and 15 dpi.Each dot in the bar plot represents one replicate, the inoculation experiments were performed with 30 plants/replicate with 6 mock-treated plants as a control.Data are presented as mean ± SD (n = 3 biological replicates for Col-0 and atx5 atxr6; n = 1 for atxr5 and atxr6).The experiments were repeated three times with similar results.c Flow cytometry assay shows that loss of MBD9 but not BRCA1 could rescue DNA re-replication phenotype in atxr5 atxr6.d Heat map shows that loss of MBD9, or SAC3B, but not BRCA1, could suppress transcriptional re-activation of TEs in atxr5 atxr6.The quantification was ), and then performed genome-wide chromatin immunoprecipitation-sequencing Fig. 3 \| Fig. 3 \| Rep hijacks RAD51 and RPA1A to facilitate viral DNA amplification.a Y2H screening pinpointed RAD51 and RPA1A as targets of Rep protein.The negative controls are AD/BD empty vectors.At least 16 independent colonies for each combination were tested and showed similar results.b Co-IP assay validated interactions of Rep with RAD51 and RPA1A in planta.FM-FVE 77 , FM-CYCB1, and Coomassie blue staining of blots serve as negative and loading controls, respectively.The experiments were repeated twice with similar results.c ChIP-qPCR assay shows the binding of RAD51 and RPA1A on viral genome.IgG is a negative control.Each dot in the bar plot represents one replicate, experiments were performed with 36 plants/replicate.Data are presented as mean ± SD (n = 3 biological replicates).d IGV file shows transcript levels of RAD51 and RPA1A in mock-treated or virusinoculated Col-0 and atxr5 atxr6 (normalized by RPKM).Scales for the distinct loci are shown on left as solid lines.e Representative phenotypes of CaLCuV-inoculated Col-0, atxr5 atxr6, rpa1a, rad51 (-/-), rad51 (+/-), rad51 (-/-) atxr5 atxr6, and rad51 (+/-) atxr5 atxr6 plants.Photographs were taken at 16 dpi.Scale bars, 1 cm.f Percentages of symptomatic plants induced by CaLCuV-inoculation in indicated backgrounds at 11, 13, and 15 dpi.g q-PCR assays show the amount of viral DNA A in CaLCuVinoculated plants indicated at 16 dpi.In (f) and (g), each dot in the bar plot represents one replicate, most experiments were performed with 36 plants/replicate except for rad51(-/-) and rad51(-/-) atxr5 atxr6 where 24 and 15 plants were used for each replicate, respectively.Data are presented as mean ± SD (n = 3 biological replicates).Normalization of viral DNA was conducted as in Fig. 2g.Statistics in Figs.3c, f, and g were performed with unpaired two-tailed student t-test, , , * and ****, P < 0.05, 0.01, 0.001, and 0.0001, respectively.Source data are provided in the Source Data File. Fig. 5 \| Fig. 5 \| Recruitment of HRR factors onto unstable genomic DNA and defense related genes prevents the loading of HRR factors on viral genome.a, b ChIP-qPCR assays show that RAD51 enrichment was significantly increased over rDNA and selected PCGs but decreased over viral DNA in atxr5 atxr6 vs Col-0 in a natural condition (a) or in a condition where increased number of symptomatic atxr5 atxr6 plants were purposedly collected (b).AT2G45820 serves as a negative control.c Overview of RAD51 signal change in atxr5 atxr6 and Col-0 upon virus inoculation.d Distributions of RAD51 and RPA1A ChIP signal (normalized with the reads of mitochondrial DNA) in different locus categories in mock-treated and virusinoculated plants.e Violin plot shows transcription changes from the loci with unchanged and increased RAD51 ChIP signal in virus-inoculated vs mock-treated atxr5 atxr6.Horizontal lines in the bar plots display the 75th, 50th and 25th percentiles, respectively.Whiskers represents the minimum and maximum values. ); 2) the viral protein Rep hijacks RAD51 and RPA1A among other DNA repair components to facilitate viral amplification (Fig. 3c-g); 3) although MBD9, SAC3B, and BRCA1 play opposite roles in regulating heterochromatin amplification and TE reactivation, loss of any of these genes interrupts activation of HRR and restores viral amplification in atxr5 atxr6 to similar levels observed in Col-0.These observations suggest that heterochromatin amplification and TE reactivation in atxr5 atxr6 do not function as a prophylactic system against CaLCuV infection (Fig. 1c-g); 4) reduced H3K27me1 causes extra copies of heterochromatic DNA, including rDNA and ncRNA loci, which induces the recruitment of HRR factors such as RAD51 and RPA1A at heterochromatic loci in the host (Fig. 4c-f); 5) deficient nYFP-Rep+-cYFP- Fig. 7 \| Fig. 7 \| BRCA1, HOP2 and CYCB1 regulate the antiviral defense in an RAD51dependent manner.a Y2H assays validated the interactions of RAD51 with CYCB1 and BRCA1.b Co-IP assay validated the interactions of RAD51 with CYCB1, BRCA1, and HOP2 in planta.FM-Rep and Coomassie blue staining of blots serve as negative and loading controls, respectively.The experiments were repeated twice with similar results.c ChIP-qPCR assay shows that BRCA1, HOP2 and CYCB1 are required for recruitment of RAD51 over rDNA and PCGs in atxr5 atxr6.d ChIP-qPCR assay shows that BRCA1, HOP2, and CYCB1 are required for reduced binding of RAD51 over the viral genome in atxr5 atxr6.e Representative pictures of the virusinoculated plants in indicated backgrounds.Photographs were taken on at 15 dpi.Scale bars, 1 cm.f Percentages of symptomatic plants of virus-inoculated plants in indicated backgrounds.g, q-PCR assays show the amount of viral DNA A in CaLCuVinoculated plants indicated at 16 dpi.Normalization of viral DNA was conducted as Fig. 2g.In (c), (d), (f), and (g), each dot in the bar plot represents one replicate,	2023-11-20T06:17:49.868Z	2023-11-18T00:00:00.000	{ "year": 2023, "sha1": "461bef3c80501ed4d0761e950564725b43a93c95", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41467-023-43311-1.pdf", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "7e3b7e7d4b2468b5578c432570a6ccece4f10c3e", "s2fieldsofstudy": [ "Biology", "Environmental Science" ], "extfieldsofstudy": [ "Medicine" ] }
6398774	pes2o/s2orc	v3-fos-license	RTB Lectin: a novel receptor-independent delivery system for lysosomal enzyme replacement therapies Enzyme replacement therapies have revolutionized patient treatment for multiple rare lysosomal storage diseases but show limited effectiveness for addressing pathologies in “hard-to-treat” organs and tissues including brain and bone. Here we investigate the plant lectin RTB as a novel carrier for human lysosomal enzymes. RTB enters mammalian cells by multiple mechanisms including both adsorptive-mediated and receptor-mediated endocytosis, and thus provides access to a broader array of organs and cells. Fusion proteins comprised of RTB and human α-L-iduronidase, the corrective enzyme for Mucopolysaccharidosis type I, were produced using a tobacco-based expression system. Fusion products retained both lectin selectivity and enzyme activity, were efficiently endocytosed into human fibroblasts, and corrected the disease phenotype of mucopolysaccharidosis patient fibroblasts in vitro. RTB-mediated delivery was independent of high-mannose and mannose-6-phosphate receptors, which are exploited for delivery of currently approved lysosomal enzyme therapeutics. Thus, the RTB carrier may support distinct in vivo pharmacodynamics with potential to address hard-to-treat tissues. Scientific RepoRts \| 5:14144 \| DOi: 10.1038/srep14144 of mutant cell lines depleted of specific pathways, RTB-directed ricin has been reported to enter cells by at least six distinct endocytic routes [4][5][6] . These include endocytosis via clathrin-dependent mechanisms, and clathrin-independent mechanisms (which account for up to 70% of fluid uptake in some cells 4 ), including pinocytosis/macropinocytosis 7 . The dominant route of RTB-mediated uptake involves absorptive-mediated rather than receptor-mediated mechanisms 8 . As all cell types tested are sensitive to ricin, albeit at different levels 3 , we propose that the nontoxic RTB lectin may be an effective carrier for delivering large glycoprotein therapeutics into a broad array of cells with significant potential to enhance disease correction in tissues that are not effectively addressed by current ERTs. As a first step in determining whether RTB can deliver corrective doses of a human ERT into disease cells and lysosomes, we produced RTB genetic fusions with human α -L-iduronidase (IDUA). IDUA is the replacement enzyme for mucopolysaccharidosis type I (MPS I, also called Hurler, Hurler-Scheie Syndrome) and is responsible for degrading glycosaminoglycans (GAG; e.g., heparan sulfate and dermatan sulfate). Severe IDUA deficiency (Hurler) leads to profound visceral, skeletal, cardiac and neurological abnormalities and early death. An FDA-approved ERT for MPS I was approved in 2003, but has not been effective in addressing the CNS-related pathologies 1,2 . Here we describe the production of human IDUA and RTB:IDUA fusions in the leaves of intact Nicotiana benthamiana plants using a transient expression system. The purified RTB:IDUA fusion product retained both lectin-binding and IDUA enzymatic activities, was efficiently taken up by human cells, and reduced GAG levels in Hurler patient fibroblasts. Results Expression of IDUA and RTB:IDUA in N. benthamiana leaves. Human IDUA has been previously produced in transgenic plants or plant cells, including Nicotiana species [9][10][11][12] . In all cases, plant-made IDUA was enzymatically active but yields were relatively low. We therefore tested two approaches for increasing IDUA product yield: codon optimization and transient expression strategies. As the native human IDUA coding sequence (wtIDUA) is very GC-rich (67%), the gene was re-synthesized based on tobacco codon preferences yielding a 44% GC sequence (optIDUA). Plant expression constructs containing wtIDUA or optIDUA, with or without an N-terminal RTB fusion partner, were developed as diagrammed in Fig. 1a. These constructs were used for Agrobacterium-mediated transient expression in leaves of intact N. benthamiana plants. The harvest time of infiltrated leaves that support greatest product yield is protein-dependent 13 . Therefore, initial studies compared product yields by Western immunoblot analysis of proteins from leaves harvested at 48, 72, 96, and 120 h after Agrobacterium infiltration of each construct. Subsequent comparisons utilized the leaves harvested at the highest yielding time-point for each construct. Western analyses showed cross-reactive product of the predicted sizes (approximately 75 kDa for IDUA and 110 kDa for the RTB:IDUA fusion protein) detected with an anti-IDUA monoclonal antibody (Fig. 1b). In contrast, extracts from leaves infiltrated with Agrobacterium bearing an "empty vector" (pBibKan) showed no cross-reactive proteins or IDUA activity. Yields of the fusion products were lower than IDUA alone. Codon-optimized versions of IDUA (optIDUA and RTB:optIDUA) provided substantial yield improvements ( Fig. 1b) and were selected for all further analyses, subsequently referred to as plant-derived IDUA (pld-IDUA) and RTB:IDUA. Characterization of RTB:IDUA fusion protein. The IDUA enzymatic and lectin-binding activities of plant-made RTB:IDUA were evaluated. Crude extracts of pld-IDUA and RTB:IDUA showed IDUA activity (Fig. 1b) indicating that the C-terminal HIS-tag and N-terminal RTB did not eliminate enzymatic function. Western analyses of the RTB:IDUA fusions revealed a breakdown product (~75 kDa) that cross-reacted with anti-IDUA antibodies, which suggests there may be a cleavage-sensitive site between the lectin and enzyme domains. To confirm that IDUA activity was associated with RTB:IDUA and not due solely to the 75 kDa product (presumably IDUA that has lost RTB), RTB:IDUA was further enriched using lectin affinity chromatography (Fig. 1c). The lactose-affinity step yielded an RTB:IDUA-enriched fraction that lacked the IDUA breakdown product but retained both lectin and IDUA enzyme activity. In order to selectively detect and quantify only lectin-active/IDUA-active RTB:IDUA in crude extracts and in purified fractions, a dual bioactivity assay was designed as a modification of the asialofetuin "functional ELISA" used to quantify RTB lectin binding. In this assay, the galactose-rich asialofetuin glycoprotein serves as the capture molecule for RTB, with a 4-MUI-based fluorescence assay subsequently used to quantify IDUA activity in lectin-active product. Comparing total IDUA activity in crude extracts with asialofetuin-bound IDUA activity (i.e. lectin-active RTB:IDUA), the asialofetuin-bound IDUA activity represents only 21% of the total value obtained in the crude plant extracts (0.08 Units/mg TSP compared to 0.4 Units/mg TSP of total IDUA activity). However, a relatively small proportion of RTB:IDUA was detected in the lactose column flow-through compared to crude fractions (Fig. 1c). This discrepancy suggests that the RTB fusion partner may reduce the specific activity of IDUA. To directly assess this, enzyme kinetics of purified RTB:IDUA (using lactose and size exclusion chromatography) were compared with mammalian cell-derived IDUA (mcd-IDUA). The identity of purified RTB:IDUA was further confirmed by mass spectrometry of tryptic peptides with 32 exclusive unique peptides identified (7 specific for RTB including the N-terminal peptide and 25 for IDUA) and representing 37% coverage (see Methods and Supplementary Fig. S2 online). Molar specific activities of the purified enzymes were Scientific RepoRts \| 5:14144 \| DOi: 10.1038/srep14144 estimated by measuring the Units (nmol of 4MU/min) generated per nmol of protein (Units/nmol). Under our assay conditions, RTB:IDUA has a V max of 622.1 ± 14.7 Units/nmol and a K m of 134 μ M, in comparison to mcd-IDUA with a V max of 897.7 ± 7.9 Units/nmol and K m of 197 μ M (Fig. 1d). These data suggest that the N-terminal placement of RTB reduces IDUA specific activity, which was observed previously with IDUA bearing an N-terminal RAP (receptor-associated protein) fusion 14 . Cellular uptake, GAG reduction and phenotype correction. To test whether RTB:IDUA is capable of delivering its enzymatic cargo to the lysosomal site of GAG substrate accumulation, lectin-active/ IDUA-active RTB:IDUA was used to treat human fibroblasts from normal and iduronidase-deficient (Hurler and Hurler-Scheie syndrome patients) individuals. In addition to RTB:IDUA, two controls samples were affinity-purified from infiltrated leaves: 1) pld-IDUA, plant-derived IDUA (purified based on the His-tag), and 2) pBIB-Kan, the plant-background negative control comprised of equivalent lactose affinity-purified fractions from leaves infiltrated with A. tumefaciens bearing a pBIB-Kan "empty vector". The final IDUA-containing products were quantified based on IDUA enzymatic activity and, unless noted, treatments of human fibroblasts were based on equivalent IDUA activity units. As shown in Fig. 2a, treatment of Hurler cells for 24 h with RTB:IDUA resulted in significant reduction in intracellular GAG content (p < 0.05) with correction equivalent to that observed with mammalian cell-derived IDUA (mcd-IDUA). In contrast, treatment of cells with "empty vector" control plant fractions (pBIB-Kan) did not reduce GAG content, indicating that no endogenous components from plants impacted GAG levels in fibroblasts. Some GAG reduction was seen in the Hurler cells treated with plant-derived IDUA. However, these levels were not significantly different when compared to the controls. It should be noted that plant-made IDUA will contain mannose-terminated glycans and thus, a low levels of uptake may be possible via high mannose receptor interactions, although this route is clearly less efficient than the adsorptive-mediated uptake conferred by RTB. Analogous experiments were subsequently performed using a Hurler-Scheie patient fibroblast line. Hurler-Sheie fibroblasts showed a higher GAG accumulation upon reaching confluence than the Hurler cell line and treatment with RTB:IDUA or mcd-IDUA GAG levels were determined in confluent cultures of normal human fibroblasts (white bar) or Hurler patient fibroblasts (black bars) that were untreated or treated for 24 h with IDUA [at 1 Unit (nmol/min) IDUA equivalents per ml of cell culture media] or "empty-vector" control fractions. mcd-IDUA, mammalian cell-derived IDUA; RTB:IDUA, plant-derived fusion protein; pld-IDUA, plant-derived IDUA; pBibKan, similarly processed protein fractions from "empty-vector" leaf extract. Each point is based on at least 3 independent culture plates (see Methods). (b) IDUA correction of lysosomal phenotype based on high throughput imaging assay. Untreated normal, untreated Hurler-Sheie fibroblasts and enzyme treated (50 ng/ ml culture medium for 24 h) Hurler-Sheie fibroblast were analyzed for LysoTracker ® signal as described in Methods. Data was expressed as percentage of LysoTracker ® signal detected relative to the total signal obtained from the untreated Hurler-Sheie fibroblast. (c) Dose response for RTB:IDUA-mediated correction of Hurler-Sheie cells. Cells were treated for 24 h at the indicated protein concentrations of mcd-IDUA or RTB:IDUA (ng purified protein/ml culture media; not adjusted to IDUA equivalents) and then analyzed based on percentage of LysoTracker ® signal per cell. (d) Kinetics of phenotype correction comparing mcd-IDUA and RTB:IDUA. Hurler-Sheie cells treated with mcd-IDUA or RTB:IDUA (50 ng/ml culture medium) were stained, fixed and analyzed at times indicated. Statistical significance vs. untreated was assessed by Dunnett's multiple comparisons test, *P < 0.005, P < 0.05. All experiments were normalized to untreated Hurler or Hurler-Sheie fibroblasts grown in the same plate and conditions to accommodate for experimentto-experiment variations in GAG levels and lysosomal phenotype are observed in disease fibroblast lines due passage number and growth rates. substantially reduced GAG levels. These results indicate that the RTB carrier delivers corrective levels of active IDUA to the intracellular sites of GAG accumulation. RTB:IDUA-mediated correction of the MPS I cellular disease phenotype was further characterized using a newly developed high-throughput imaging assay that captures lysosome area and number based on staining with LysoTracker ® red (see Methods). Image data from treated or untreated cells were analyzed based on pixels/cell (total LysoTracker ® red pixels per image divided by cell number based on DAPI-stained nuclei) under fixed magnification and capture parameters. This assay effectively distinguished normal and Hurler-Scheie fibroblasts, with Hurler-Scheie fibroblasts showing significantly higher lysosomal signal (80% higher red pixels/cell) consistent with extensive GAG accumulation (Fig. 2b). Treatment of Hurler-Scheie fibroblasts with 50 ng/ml of the IDUA-containing recombinant proteins for 24 h, resulted in signal reduction from 100% to about 40% (Fig. 2b). This method was used to compare the kinetics of lysosomal phenotype corrections of mcd-IDUA and RTB:IDUA under varying doses (Fig. 2c) and times of treatment (Fig. 2d). Following 24 h incubation with different concentrations of recombinant protein, RTB:IDUA demonstrated similar efficacy for phenotype correction of the Hurler-Sheie fibroblast compared to mcd-IDUA (Fig. 2c). It should be noted that equivalent amounts (ng) of purified protein were used in this assay as opposed to molar equivalents, suggesting that RTB:IDUA may actually be more effective than mcd-IDUA under these conditions even with the lower apparent molar V max . The kinetics of lysosomal correction (Fig. 2d) were similar for Hurler-Sheie fibroblasts treated with either RTB:IDUA or mcd-IDUA (50 ng/ml). RTB:IDUA uptake is independent of high mannose receptor -and mannose-6-phosphate receptor-mediated routes. We propose that RTB delivers lysosomal enzymes by fundamentally different mechanisms than currently approved lysosomal ERTs which exploit either high-mannose (MMR) or mannose-6-phosphate (M6PR) receptors to gain access to cells and lysosomes. Unlike mammalian cells, plant cells do not use mannose-6-phosphate-terminated glycans to direct enzymes to vacuoles (plant lysosomal compartment) and do not possess the enzymatic machinery to make this modification 15 . Thus, the results described above suggest that RTB:IDUA accesses cells and lysosomes by M6PR-independent routes. The N-linked glycans of plant-made glycoproteins include mannose-terminated forms which could contribute to uptake of RTB:IDUA and plant IDUA by a high mannose receptor (MMR) route 16,17 . In order to assess a potential role of either M6PR or MMR in the internalization of plant-derived RTB:IDUA, uptake assays were performed in the presence of the respective competitive inhibitors, M6P sodium salt and mannan (Fig. 3a). Saturation of the receptors was achieved by incubating the Comparison of enzyme treatments were assessed at 24 h using the LysoTracker ® high-throughput assay outlined in Fig. 2 with statistical significance of treated vs. untreated samples assessed by Dunnett's multiple comparisons test (*P < 0.005, P < 0.05). cell cultures for two hours prior to enzyme treatment with doses of inhibitors reported to block 95% of uptake 18 . Cells were subsequently treated with the various IDUA enzymes for 24 h. While the presence of mannose-6-phosphate inhibitor clearly blocked the ability of mcd-IDUA to reduce GAG substrate levels in Hurler fibroblasts, the inhibitor had no significant effect on RTB:IDUA's ability to reduce GAG substrate content in cells (P < 0.005). We also tested RTB:IDUA-mediated uptake and GAG reduction in the presence of inhibitory levels of mannan to determine if RTB-mediated intracellular delivery was based on MMR interactions. As shown in Fig. 3a, RTB:IDUA in the presence of mannan effectively reduced GAG substrate in Hurler fibroblasts to levels similar to those attained from cells treated with either mcd-IDUA or RTB:IDUA in the absence of inhibitor (P < 0.005). We corroborated M6P-independent uptake by measuring phenotypical changes in the cell lines using the high-throughput imaging assay described above. In contrast to that observed in the absence of inhibitors, mcd-IDUA was unable to reduce the LysoTracker ® signal in Hurler-Sheie fibroblast when M6P receptors were blocked (Fig. 3b). However, M6P receptor saturation did not impede the ability of RTB:IDUA to reduce the LysoTracker ® signal in disease fibroblasts to levels obtained in the absence of inhibitors. To further characterize the role of RTB lectin activity in the uptake of IDUA, polyclonal antibodies to RTB (shown to be partially neutralizing; Liu, 2006, etd-12202006-220049) were used to block the sugar binding activity of RTB. Hurler-Sheie fibroblasts were treated with RTB:IDUA that had been pre-incubated with anti-RTB antibodies. These cells showed no reduction in LysoTracker ® signal (i.e., no lysosomal phenotype correction) indicating that uptake and/or trafficking of RTB:IDUA was blocked (Fig. 3b). In contrast, anti-RTB antibodies did not prevent mcd-IDUA-mediated correction of the lysosomal phenotype of disease fibroblasts. These results suggest that the dominant route(s) of RTB-mediated uptake is independent of either M6PR or MMR but is dependent on sugar-binding and lectin-mediated endocytosis. Discussion In summary, we have demonstrated three main points. First, plants are capable of producing large bioactive glycoprotein fusion products comprising an RTB lectin and human lysosomal enzyme that retain both lectin selectivity and α -L-iduronidase enzyme activity. Second, the RTB:IDUA fusion product effectively reduces the GAG disease substrate and corrects the lysosomal phenotype of Hurler and Hurler-Scheie patient fibroblasts with kinetics equivalent to mammalian-cell-derived IDUA. This data represents the first demonstration that the plant lectin RTB delivers fully functional enzymes into cells and into the critical subcellular sites of lysosomal disease substrate accumulation. Third, RTB:IDUA corrects the disease phenotype of Hurler and Hurler-Scheie fibroblasts by mechanisms that are independent of either high-mannose receptors (MMR) and M6P receptors suggesting the potential for distinct in vivo pharmacodynamics compared to current ERTs. All current FDA-approved LSD enzyme replacement therapeutics rely on one of only two mechanisms to carry corrective enzyme into cells and to the lysosomal sites of disease substrate accumulation. ERTs for Gaucher disease utilize glucocerebrosidase modified to display terminal mannose residues on their N-linked glycans. This modification directs cell uptake via the high mannose receptors (MMR), which are particularly abundant on macrophages and other cells of monocyte lineage -cells that show significant disease involvement in Gaucher patients. Currently approved ERTs for Fabry, Pompe and mucopolysaccharidosis types I, II, IV A and VI diseases use recombinant enzymes that display N-linked glycans with terminal mannose-6-phosphate residues targeting delivery based on interactions with the M6P receptors. M6P receptors function primarily within the endomembrane system to facilitate delivery of acid hydrolases to the lysosome, especially in mediating trans-Golgi to early/late endosome transport 19,20 . M6P-mediated ERT uptake into cells exploits the M6PR salvage pathway that recovers M6PR from the plasma membrane. Although ERTs have been highly beneficial to patients with these diseases, the ineffective correction of neurological, ocular and skeletal disease manifestations remains a major drawback of current therapies dependent upon MMR-and M6PR-dependent delivery. M6P transport to the brain has been demonstrated to be developmentally regulated in early postnatal life and lost in adulthood 21 . Thus, there is keen interest in developing new enzyme delivery strategies with potential to more effectively target a broader array of cells and tissues. In the current study, we showed that RTB effectively delivers associated human lysosomal enzymes into cells and lysosomes by mechanisms that are independent of either MMR or M6PR. Treatment of Hurler and Hurler-Scheie fibroblasts with RTB:IDUA resulted in reduction of cellular GAG levels to "normal" levels at doses and with correction kinetics equivalent to mammalian cell-derived IDUA. The dominant routes of RTB-mediated uptake into mammalian cells involved absorptive-mediated endocytosis based on lectin interactions with the galactose/galactosamine-containing glycoproteins and glycolipids abundant on mammalian cell surfaces 22,23 . In contrast to receptor-mediated uptake, adsorptive-mediated endocytosis is considered less-saturable due to the wider interaction with non-specific moieties expressed at the luminal surface of cells 24 . Using these mechanisms, various plant lectins have been shown to facilitate drug delivery to many of the "hard-to-treat" organs (e.g., eye, lung and brain 25,26 , reviewed in 27 ). Historically, RTB-mediated uptake based on ricin cytotoxicity has been demonstrated in essentially all of the tissues/cell types considered challenging for therapeutic glycoproteins delivery including peripheral and central neurons, brain, bone, lung and heart [28][29][30] . We therefore anticipate that RTB will provide greater enzyme delivery to these challenging tissues than ERTs that rely exclusively on MMR and M6PR interactions. We produced RTB:IDUA using an Agrobacterium-mediated transient expression system in the leaves of Nicotiana benthamiana. Production of the non-toxic RTB lectin in a heterologous plant system eliminates any possible ricin toxicity (no RTA is present). In addition, plant-based production systems may provide advantages in safety (do not support human viruses), cost (especially initial capitalization costs), and scalability compared to mammalian-cell-based systems 31,32 . The ability of plant systems to provide clinically efficacious ERTs was demonstrated with FDA approval of glucocerebrosidase (Elelyso) produced in carrot cells for treatment of Gaucher disease. Clinical trials showed no difference in effectiveness or immunogenicity between plant-and animal-cell-derived products 33,34 . The plant-based Gaucher ERT has been safely administered to patients since 2007 35 , reducing therapy cost and addressing safety and supply shortfalls with the current CHO-based product. Although glycoproteins made using the N. benthamiana platform have not yet attained FDA approval for human use, monoclonal antibodies (e.g., ZMapp antibodies used to treat Ebola) and virus-like particles made using this system are in clinical evaluation and show efficacy in primates and humans 36,37 . That plants correctly synthesize and assemble fully functional antibodies underscores their potential to produce highly complex glycoproteins. In this study, we produced a 110 kDa fusion protein that placed the RTB plant lectin (34 kDa) at the N-terminus of human IDUA (~75 kDa). Active IDUA had been produced previously in plants 9-11 but at relative low levels. We found that using a "codon-optimized" version of IDUA based on tobacco codon preferences provided greater yields in plants, especially of the large RTB:IDUA fusion (Fig. 1b). Enzyme kinetic analyses comparing RTB:IDUA to mammalian cell-derived IDUA showed a lower V max for the plant-derived fusion proteins (based on molar equivalents) although the enzymes behaved equivalently in the disease correction assays with MPS I fibroblasts (Fig. 2c). It may be that placement of the lectin partner at the N-terminus of IDUA negatively impacts IDUA conformation or substrate access. Consistent with this interpretation, previous studies showed that addition of an unrelated peptide to the N-terminus of IDUA also resulted in reduced IDUA enzyme kinetics 14 . To further test impacts of fusion orientation, we recently produced fusions that placed RTB at the C-terminus of IDUA (IDUA:RTB). Preliminary analyses indicate that this IDUA:RTB fusion provided IDUA activity and kinetic values equivalent to mcd-IDUA (Condori, Acosta and Cramer, unpublished results). Lysosomal storage diseases represent a devastating group of rare diseases and ERT therapies have revolutionized patient care in those diseases where they are available. New delivery strategies are needed to address the complex clinical presentations of these diseases and to mobilize corrective enzyme to sites such as brain and bone that are not treated by the current generation of ERTs. The research presented here introduces a new ERT carrier system based on the plant RTB lectin that shows efficient delivery of corrective enzyme to disease cells and lysosomes by receptor-independent routes. RTB-mediated delivery mechanisms are fundamentally different than those used by current ERT drugs. The RTB carrier may therefore provide novel biodistribution and pharmacological behavior to associated ERTs in vivo. Evaluation of the RTB:IDUA ERT in the MPS I mouse model 38,39 will be required to determine whether RTB improves ERT access and disease correction in these "hard-to-treat" tissues. If successful in this LSD model, RTB-mediated delivery may have significant implication for therapeutic protein delivery impacting a broad spectrum of diseases. Construct design and Agrobacterium-mediated transient expression. Construction of the cassette containing the double enhanced 35S constitutive plant promoter 40 , a translational enhancer from Tobacco Etch Virus (TEV) 41 , the plant signal peptide derived from the potato patatin tuber storage protein (pat) 42 , and the RTB sequence was described previously (Reidy, 2006, etd-12202006-220049). For the fusion constructs, the RTB sequence was PCR-amplified using Pfu DNA polymerase to generate the RTB coding fragment. Sequences encoding human IDUA were synthesized (GeneArt) based on GenBank M74715.1 in two different versions; one utilized the human DNA sequence; the other was codon-optimized based on Nicotiana tabacum codon preferences and GeneArt's expression optimization algorithm. Sequence encoding a C-terminal 6X-histidine tag was added to facilitate detection and purification. IDUA-His and RTB:IDUA expression cassettes (diagrammed in Fig. 1a) were cloned into the plant expression/transformation vector pBIB-Kan 43 using In-phusion polymerase dry-down PCR Cloning Kit (Clontech). PCR and In-phusion cloned products were confirmed by DNA sequence analyses. Plasmids containing the various gene constructs were introduced into Agrobacterium tumefaciens strain LBA4404 using a freeze/thaw method 44 . A. tumefaciens-mediated transient expression of these constructs in Nicotiana benthamiana plants was performed as described previously 13 . Extraction and quantification of recombinant protein. Plant tissue was ground in a mortar and pestle under liquid nitrogen and then homogenized with extraction buffer (100 mM Tris-HCl, 150 mM MgCl 2 , 10 mM Na 2 S 2 O 5, 2 mM PMSF, and 0.1% Tween 20; pH 7.5). Samples were centrifuged at 13,000 × g for 30 min at 4 °C and the clarified supernatant was used in Western immunoblot analysis, enzymatic activity assays, and lactose binding analysis. Scientific RepoRts \| 5:14144 \| DOi: 10.1038/srep14144 Western immunoblot analysis. Proteins from crude extracts were fractionated on 12% Novex ® NuPAGE Bis-Tris polyacrylamide gel (LifeTechnologies); by SDS-PAGE along with the Dual Color Precision Plus Protein Standard (BioRad). Proteins were transferred to nitrocellulose membrane and immuno-detected using the SNAPi.d. Protein Detection System (EMD Millipore). Membranes were blocked with 3% BSA in PBS + 0.05% Tween-20. A monoclonal anti-IDUA (R&D Systems) was used as the primary antibody and detected using an affinity-purified alkaline phosphatase-conjugated goat anti-mouse secondary antibody (BioRad). Immunoblots were developed with CDP-Star (Roche) chemilluminescent substrate and images captured on film in accordance with manufacturer instructions. Enzymatic activity. IDUA enzymatic activity was determined using the fluorogenic substrate 4-methylumbelliferyl-α -iduronide (4MUI) (Santa Cruz Biotechnology). A previously reported protocol for measuring IDUA activity in plant extracts was adapted 9,11 . Briefly, 10 μ l of substrate solution (0.75 mM 4MUI in assay buffer (0.1 M sodium acetate pH 4.8, 1 mM sodium metabisulphite, 3.5 mg/ml BSA)) was mixed with 5 μ l plant crude extract in a black-walled 96-well microtiter plate and incubated in the dark at 37 °C for 60 min. Corresponding extracts from "empty vector" control plants were used to establish background. A 4-methylumbelliferone standard curve was generated by serial dilutions ranging from 0.23-15 μ M. Reactions were terminated by adding a pH 10.7 stop solution and fluorescence was detected (excitation = 355 nm, emission = 460 nm). One unit of enzyme activity is defined as 1 nmol of 4-MU produced per min at 37 °C. Dual bioactivity assay. An assay was developed to quantify RTB:IDUA fusion proteins that takes into account both RTB and IDUA activities. It is based on lectin-mediated affinity to glycoproteins by RTB and direct IDUA activity using 4MUI. Immulon 4HBX plates were coated with asialofetuin (Sigma-Aldrich) containing solution (300 μ g/mL) in bicarbonate buffer (pH 9.5) followed by incubation at room temperature for 1 h. Following a washing step with 0.5% Tween-20 in PBS, 100 μ L of diluted samples (1:20) in PBS + 1% BSA were added to the asialofetuin-coated plate and incubated for 1 h at room temperature. Following a second wash step, an aliquot of extracts from plants infiltrated with an empty vector (pBIB-Kan; 5 μ l) was added to all test sample wells to control for components that may be present in plant crude extracts. A control crude sample containing an aliquot of the recombinant fusion crude extract (5 μ l) was applied to unloaded wells in order to determine the activity emitted from any enzymatically active IDUA (with or without lectin binding activity). Reaction was initiated with 10 μ l of substrate solution (0.1 M sodium acetate pH 4.8, 1 mM sodium metabisulphite, 3.5 mg/ml BSA and 0.75 mM 4MUI) and incubated for 60 min at 37 °C. Fluorometric determination for detecting and quantifying fluorescent breakdown of the substrate was performed as described above for IDUA enzymatic assays. Protein purification, quantification and confirmation of identity. RTB:IDUA fusion protein and plant-synthesized IDUA control protein were extracted and enriched using affinity chromatography methods; RTB:IDUA was purified based on lectin affinity using lactose resins (Sigma) while the HIS-tagged IDUA (pld-IDUA) was purified by IMAC affinity chromatography using a nickel-based Ni-superflow ® resin (Clontech). Tissue was ground in liquid nitrogen and resuspended in 100 mM Tris-HCl, pH 7.5, 150 mM MgCl 2 , 10 mM Na 2 S 2 O 5, 2 mM PMSF, and 0.1% Tween 20 at a 1:2 (w/v) ratio. Samples were clarified by centrifugation at 20,000 × g for 20 min. Supernatant of RTB:IDUA-containing extracts were incubated with α -Lactose-agarose resin for 2 h at room temperature prior to transfer to a poly-prep chromatography column (BioRad). Resin was washed with 5X column volumes of 100 mM Tris-HCl buffer, pH 7.5. RTB:IDUA was eluted with wash buffer containing 500 mM D-galactose. For HIS-tagged plant-derived IDUA, the protein was purified on Ni-Superflow ® resin as described in the Clontech user manual. Elution fractions were concentrated using Amicon Ultra-4 centrifugal filters (30,000 MWCO; EMD Millipore). Buffer-exchange with 100 mM Tris-HCl pH 7.0 (for RTB:IDUA) or 50 mM sodium acetate pH 4.8, 500 mM NaCl (for pld-IDUA) was carried out using Zeba TM spin desalting columns, 7,000 MWCO (Thermo). Final fractions were sterile-filtered using a 0.2 μ m centrifugal filter. Tissue from plants infiltrated with A. tumefaciens containing an empty vector (pBibKan) was extracted and processed following the lactose affinity procedure to serve as a corresponding negative control. Quantification of the activity units for purified plant-produced products and for mammalian cell-derived mcd-IDUA (R&D Systems) was determined by the IDUA enzymatic activity assay described above. In order to quantify the specific activity and kinetics of RTB:IDUA, the lactose purification fraction was further purified using size exclusion chromatography as described in manufacture manual (Hiload TM 26/60 Superdex TM 200, General Electric). Purified protein (1 μ g) was resolved by SDS-PAGE and Coomassie-stained following manufacturer's (Thermo) instructions (shown in Supplementary Fig. 1S online). The 120 kDa band was excised and analyzed at University of Arkansas Medical Science Proteomics Core by in-gel trypsin digestion and tandem mass spectrometry (MS/MS) ( Supplementary Fig. S2 online). Enzyme kinetic analyses. Purified mcd-IDUA (R&D Systems) and RTB:IDUA, (99% and 98% purity, respectively) were assessed under the same assay conditions. RTB:IDUA was further purified using size Scientific RepoRts \| 5:14144 \| DOi: 10.1038/srep14144 exclusion chromatography and purity was quantified by BioRad Experion ® Pro260 chip ( Supplementary Fig. S1 online). Concentration was calculated from A280 measurements and theoretical extinction coefficient of 183,075 ∈ M −1 cm −1 . RTB:IDUA and mcd-IDUA were diluted to 2 nM in assay buffer. Substrate (4MUI) was diluted in assay buffer to concentrations ranging from 2000 to 15.6 μ M. Equal volumes (10 μ L) of protein and substrate solution were combined to initiate the reaction and incubated as described above. Reactions were terminated by the addition of 280 μ l of stop solution. Results were plotted as units of IDUA activity per min for each substrate concentration. Data analysis was performed using GraphPad Prism software. Cellular uptake and GAG quantification assay. A normal fibroblast (GM00010) cell line and two diseased cell lines -Hurler fibroblasts (GM01391) and Hurler-Scheie fibroblasts (GM00963) were purchased from Coriell Cell Repositories and cultured in 150 mm plastic dishes in conditions recommended by the provider. For experiments, culture media of confluent cells was replaced with serum-free MEM containing varying test doses of recombinant protein, control protein (purified protein fraction from plants transformed with an empty vector) or no treatment. For receptor inhibition studies, media containing 4 mM mannose-6-phosphate sodium salt or 4 mg/ml mannan (Sigma) was added 2 h prior to addition of various IDUA proteins or control samples. At treatment initiation, proteins stocks were added directly to the inhibitor-containing media and cells were incubated for 24 hours at 37 °C, 5% CO 2 . Following incubation, cells were harvested using Trypsin-EDTA, washed twice with PBS after centrifugation (1000 × g, 4 °C, 5 min), and cell pellets stored at -20 °C for GAG quantification assays. The glycosaminoglycan (GAG) content in fibroblasts was quantified by measuring the absorbance of the complex formed by sulfated GAGs and dimethylmethylene blue (DMMB) as described elsewhere 45 . Fibroblast responses to the various treatments were analyzed using Dunnett's multiple comparisons test in GraphPad Prism software. High throughput imaging assays. Hurler, Hurler-Scheie, and normal fibroblasts were plated at 10 5 cells/ml in a black-walled, clear bottom 96-well plate. Following a 3 h period to allow cell attachment, media was replaced with 100 μ l serum-free media containing the various test treatments. For M6P receptor inhibition studies, 75 μ l of media containing 5.3 mM M6P sodium salt was added 2 h prior to addition of various IDUA variants or control samples; at treatment time, proteins were diluted in 25 μ l of media and added to each well yielding a final concentration of 4 mM M6P. Cells were incubated for 24 h at 37 °C and 5% CO 2 . Following the 24 h treatment period, cells were treated with 600 nM LysoTracker ® red (Invitrogen) for 20 min and fixed with 4% paraformaldehyde (8 min). Cells were counterstained with DAPI (Invitrogen) and analyzed using a BD Pathway 855 High Content Bioimager (BD Biosciences). Images for each well/repetition were taken using a 2 × 2 montage with the 20X objective yielding an average of 200 cells per image. Fluorescent signal (Excitation/Emission) at 560/645 (red) and 380/435 (blue) was acquired using the same exposure and laser autofocus parameters for each well. Image data acquisition was obtained using the Attovision ® software analysis tools (BD Biosciences). Cell count was obtained using polygon segmentation for the nucleus blue signal. LysoTracker ® regions of interest (ROI) were obtained by polygon segmentation of one of each dots detected at the red signal. All images within an experiment were analyzed using the same image processing and segmentation parameters. Data analysis was performed using BD Data explorer ® software (BD Biosciences). An average of LysoTracker ® pixels per cell were calculated for each image. Averages of at least fifteen images (n = 15) were used to estimate the value of each treatment. Plate-to-plate variation was observed depending on the uptake efficiency of the LysoTracker ® reagent, stability of the probe or exposure value used for each plate. Due to this inherent variation, untreated disease cells and normal cells were run as controls for each plate. To compare across plates, data was expressed based on the LysoTracker ® signal of each treatment as a percentage of the total signal obtained with untreated disease cells. Fibroblast responses to the various treatments were analyzed using Dunnett's multiple comparisons test in GraphPad Prism software.	2016-05-04T20:20:58.661Z	2015-09-18T00:00:00.000	{ "year": 2015, "sha1": "2f3dcfbfeaf89975f94f396310cacaacc4d7157e", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/srep14144.pdf", "oa_status": "GOLD", "pdf_src": "SpringerNature", "pdf_hash": "cd63d48a6f085c53c2f2fe662ef35281282cd356", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine", "Biology" ] }
841616	pes2o/s2orc	v3-fos-license	Antiarthritic effect of aqueous and ethanolic leaf extracts of Pistia stratiotes in adjuvant-induced arthritis in Sprague-Dawley rats Background Pistia stratiotes has been used effectively to treat a number of inflammatory conditions. This study aims to determine the antiarthritic effect of aqueous and ethanolic leaf extracts of P. stratiotes. Methods Arthritis was induced in Sprague-Dawley rats, paw swelling was measured, and arthritis indices were estimated in rats treated with aqueous and ethanolic leaf extracts of P. stratiotes (AQ PSE and ET PSE, respectively), methotrexate, diclofenac, dexamethasone, and normal saline-treated rats. Radiologic imaging, hematological assessment of red and white blood cells, C-reactive protein and erythrocyte sedimentation rate, as well as histopathological studies were also done. The data were analyzed using GraphPad Prism 5. Results The 30, 100, and 300 mg/kg doses of AQ PSE and the 30 and 100 mg/kg doses of ET PSE caused a significant (P ≤ 0.05–0.001) reduction in ipsilateral paw swelling, similar to the effects of methotrexate, dexamethasone, and diclofenac. Only the 30 mg/kg dose of AQ PSE caused a significant (P ≤ 0.01) reduction in contralateral paw swelling. Arthritic indices reduced significantly (P ≤ 0.05–0.001) at all drug doses, except for the 100 and 300 mg/kg doses of ET PSE. White blood cell levels decreased significantly (P ≤ 0.05–0.01) in arthritic rats treated with the 30 mg/kg dose of AQ PSE and those treated with methotrexate. Erythrocyte sedimentation rate and C-reactive protein levels were significantly (P ≤ 0.01–0.001) lower in all the treatment groups except for the rats treated with AQ PSE 300 mg/kg and ET PSE 100 and 300 mg/kg doses. The arthritic animals treated with 30 mg/kg of the aqueous extract showed no inflammatory changes in the ipsilateral paw, while the contralateral paw showed only foci of mild chronic inflammatory changes, as seen with the reference drug treatment in histopathological studies. Conclusion This study establishes that aqueous and ethanolic extracts of P. stratiotes have antiarthritic activity in Sprague-Dawley rats with induced arthritis. The aqueous extract had better activity than the ethanolic extract. Introduction Inflammation is implicated in many diseases, and its consequences tend to be worse if resolution fails and inflammation progresses to a chronic stage. 1 Although inflammation is a body defense mechanism, its role in the maintenance and aggravation of various disease conditions (such as rheumatoid arthritis) is of public health importance and makes it a subject of scientific interest and clinical concern. 2 Arthritis, an autoimmune disorder, is a chronic inflammatory disease which manifests itself in multiple joints of the body. The inflammatory process primarily affects the lining of the joints (synovial membrane), but can also affect other organs. The inflamed synovium leads to erosions of the cartilage and bone, and sometimes to joint deformity. Pain, swelling, and redness are common joint manifestations. 3 Arthritis causes disability, compromised quality of life, and premature mortality. 4 It affects approximately 0.5%-1.0% of the global adult population, with an estimated annual incidence of 12.0-24.5 males and 23.9-54.0 females per 100,000. 3,5,6 The incidence is largely consistent racially and geographically, and the peak age of onset lies between the ages of 45 and 65 years. The economic cost of arthriticrelated diseases is estimated to consume about 3% of gross domestic product in developed countries, and is somewhat lower in developing countries due to lower life expectancy. 7,8 The present decade has therefore been declared the "Bone and Joint Decade" by the World Health Organization, 9 in order to further our understanding of the impact of musculoskeletal diseases on society and individuals and to assist advancement on this front. It is against this background that we attempted to assess the antiarthritic effect of aqueous and ethanolic leaf extracts of Pistia stratiotes in a rodent adjuvant-induced arthritis model to ascertain its importance in the traditional management of inflammatory disorders. 10,11 Materials and methods Plant collection P. stratiotes was collected from the Fosu Lagoon, Cape Coast, in the Central Region of Ghana (5°7′ N and 1°16′ W) in December 2010. It was identified and authenticated by Mr GH Sam of the Department of Herbal Medicine, Kwame Nkrumah University of Science and Technology, where a voucher specimen bearing the number KNUST/ HM1/11/W002 has been deposited at the herbarium for future reference. Preparation of extracts The leaves of P. stratiotes were washed thoroughly with tap water and sun-dried. The dry leaves were milled into coarse powder by a hammer mill (Schutte Buffalo, New York, NY). In preparing the aqueous leaf extract of P. stratiotes, 700 g of the leaf powder was mixed with one liter of water. The mixture was maintained at 80°C (in a round-bottomed flask fitted with a reflux condenser) in a thermostatically controlled water bath for 24 hours and then filtered. The filtrate was freeze-dried with a Hull freeze-dryer/ lyophilizer 140 SQ FT (model 140FS275C; Hull, Warminster, PA) into powder (yield 4.7%) and stored at a temperature of 4°C in a refrigerator. This powder was reconstituted in normal saline to the desired concentration and labeled as the aqueous extract of P. stratiotes (AQ PSE) for dosing in this study. Similarly, 700 g of the leaf powder was soaked with one liter of 70% ethanol at room temperature (27°C-29°C) for 72 hours and filtered. The filtrate obtained was freeze-dried into powder (yield 5.2%). Quantities of this powder were reconstituted in normal saline at the desired concentrations to be referred to and used in this study as the ethanolic leaf extract of P. stratiotes (ET PSE). Drugs and chemicals Complete Freund's adjuvant (CFA) was a 5 mg/mL suspension of heat-killed Mycobacterium tuberculosis (strains C, DT and PN [mixed] obtained from the Ministry of Agriculture, Fisheries and Food, UK) triturated in paraffin oil (Ernest Chemist, Accra, Ghana) was used for the induction of arthritis. Diclofenac sodium (KRKA, Novo mesto, Slovenia), dexamethasone sodium (Anhui Medihel Co, Ltd, Hefei, Anhui, China), and methotrexate sodium (Dabur Pharma, New Delhi, India) were the reference anti-inflammatory agents in this study. Preparation of reference drugs The reference anti-inflammatory drugs were dissolved in normal saline for the study. The drugs were freshly prepared and administered in volumes not exceeding 10 mL/kg. Animals Male and female 6-8-week-old Sprague-Dawley rats (weight 180-200 g) purchased from the Centre for Scientific Research into Plant Medicine, Mampong-Akwapim, Ghana, were maintained in the Animal House of Department of Pharmacology, Kwame Nkrumah University of Science and Technology, Ghana. The animals were housed in polyacrylic cages (34 cm × 47 cm × 18 cm) with soft wood shavings as bedding, under ambient laboratory conditions (temperature 28°C ± 2°C, relative humidity 60%-70%, and a normal light-dark cycle). None of the females were pregnant. They were fed on a normal commercial pellet diet (GAFCO, Tema, Ghana) and had access to water ad libitum. All procedures and techniques used in these studies were in accordance with the National Institute of Health for the Care and Use of Laboratory Animals. 12 Protocols for the study were approved by the departmental ethics committee. submit your manuscript \| www.dovepress.com Preliminary phytochemical screening Screening was performed on AQ PSE and ET PSE to ascertain the presence of phytochemicals using standard procedures described by Wagner and Bladt, 13 Glasl, 14 Harborne, 15 and Kujur et al. 16 induction of arthritis Arthritis was induced as previously described by Pearson,17 with slight modification. In this process, the initial hind paw volumes (both left and right) of the experimental animals were measured by water displacement plethysmography. 18 The right hind paw of each animal was then injected with 0.1 mL of CFA. The paw volumes for both the injected paw (ipsilateral) and the noninjected paw (contralateral) were measured on day 1 after injection into the paw, and every other day. Unilateral inflammatory edema of the ipsilateral paw peaking at around days 4-6 was indicative of successful induction of adjuvant arthritis. Experimental procedure Nine days after adjuvant-induced arthritis, the animals were put into ten groups (A-J) of five and were randomized to treatment with either 30, 100, or 300 mg/kg of AQ PSE or ET PSE (orally daily), 0.3 mg/kg methotrexate (intraperitoneally every 4 days), 0.43 mg/kg diclofenac (intraperitoneally daily), 1 mg/kg dexamethasone (intraperitoneally, every other day), or 1 mL/kg normal saline (orally daily) over the experimental period. A nonarthritic control group consisting of five animals in which incomplete arthritis was induced by intraplantar injection of 0.1 mL of sterile paraffin oil (incomplete Freund's adjuvant, [IFA]) and a normal control group in which there was no induction of arthritis were also studied. The effect of the reference and test drugs on the edema component of arthritis was quantified by measuring the difference in paw volume prior to induction of arthritis (day 0) and that at the various time points of assessment up to day 27. The arthritic index for the ipsilateral and contralateral paw volumes was individually calculated and expressed as a percentage change with respect to values at day 0, and then averaged for each treatment group. Initial body weights of the rats were recorded at day 0 and on day 28. On day 28, the experimental animals were sent to the radiology unit of the Kwame Nkrumah University of Science and Technology Hospital where they were anesthetized with ketamine hydrochloride for X-ray imaging using a conventional X-ray machine (Philips, Eindhoven, The Netherlands) and industrial X-ray film (Fuji Photo Film, Tokyo, Japan). The X-ray apparatus operated at a 52 kV peak and 10 seconds of exposure with a 45 cm tube-to-film distance for lateral projections. The rats were sacrificed and blood samples were collected into MediPlus K3 ethylenediamine tetra-acetic acid tubes (Sunphoria Co, Ltd, Taipei, Taiwan) for hematological analysis using the KX-21 N automated hematology analyzer (Sysmex Corporation, Chuo-ku, Kobe, Japan) and into trisodium citrate erythrocyte sedimentation rate tubes (Chengdu Rich Science Industry Co, Ltd, Sichuan, China) for estimation of erythrocyte sedimentation rate using the Westergren method. 19 Blood was also collected into glass tubes and centrifuged (temperature 25°C, speed 4000 g) for 5 minutes using a Mikro 220R (Hettich Zentrifuge, Tuttlingen, Germany) machine to obtain the plasma, which was used to estimate C-reactive protein levels using an enzyme-linked immunosorbent assay kit (Fortress Diagnostic Ltd, Antrim, Northern Ireland). The right and left hind paws of the animals were amputated and fixed in 4% phosphate-buffered paraformaldehyde, decalcified with 1% ethylenediamine tetra-acetic acid, and embedded in paraffin. Sections were stained with hematoxylin and eosin, 20,21 and fixed on glass slides for microscopic examination at the Pathology Department of the Komfo Anokye Teaching Hospital. Data obtained for the arthritic and radiological indices as well as histopathological, hematological, and physical profiles were analyzed. Arthritic index Photographs of the arthritic rats were taken on day 28 with a camcorder (Everio™ model GZ-MG1300; JVC, Tokyo, Japan). Inflammation in each paw was graded blindly by the same person for all rats on day 28 according to the extent of erythema and edema of the periarticular tissues, using a scale of 0-4 ( Table 1). The arthritis score for each rat on day 0 was determined to be 0. The scores for each paw were then added to get the total arthritis score (maximum possible score 16 per animal), and designated as the arthritic index. radiological index Using the radiographs, the severity of bone and joint destruction was scored blindly by the same person for each hind Table 1 Grading scale for arthritic index 22,23 Score Description Unequivocal inflammation of one joint of the paw 2 Unequivocal inflammation of at least two joints of the paw or moderate inflammation of one joint 3 Severe inflammation of one or more joints 4 Maximum inflammation of one or more joints in the paw submit your manuscript \| www.dovepress.com Dovepress Dovepress limb according to the method described by Hoffmann et al. 24 Briefly, radiographic scoring was performed by assessing soft tissue swelling, periosteal new bone formation, joint space narrowing, periarticular osteoporosis, and bone destruction on a scale of 0 (normal) to 3 (maximum) per hind limb. The maximum radiographic score was 6 per animal. The radiological score for normal control rats was determined to be 0. The radiological score was termed the radiological index. Statistical analysis Results were analyzed using one-way analysis of variance followed by Dunnett's multiple comparisons test using GraphPad Prism (version 5.03; GraphPad, La Jolla, CA). Values were expressed as the mean ± standard error of the mean. P # 0.05 was considered to be statistically significant. Results The results of preliminary phytochemical screening are shown in Table 2. Intraplantar injection of CFA induced an inflammatory response characterized by paw swelling in both the ipsilateral and the contralateral paws. The response of the injected paw was biphasic, consisting of acute and polyarthritic phases corresponding to days 0-9 and 10-27 following inoculation, respectively. The acute phase response was characterized by unilateral inflammatory edema of the ipsilateral paw, peaking around days 4-6, followed by a polyarthritic phase response which began around day 9, characterized by inflammatory edema of the contralateral paw. AQ PSE 30, 100, and 300 mg/kg caused a significant reduction (P # 0.01-0.001) in ipsilateral paw thickness in the CFA-induced arthritic rats, but only AQ PSE 30 mg/kg caused a significant reduction (P # 0.01) in contralateral paw thickness ( Figure 1A and B). ET PSE 30 and 100 mg/kg achieved a significant (P # 0.05) reduction in ipsilateral paw thickness, but there was no significant (P . 0.05) reduction in contralateral paw thickness at any dose level (Figure 2A and B). Methotrexate, dexamethasone, and diclofenac caused a significant (P # 0.05-0.001) reduction in ipsilateral and contralateral paw thickness in arthritic rats ( Figure 3A and B). Compared with the arthritic animals, there was no significant increase in ipsilateral and contralateral paw thicknesses in animals with IFA-induced arthritis. Normal animals kept under experimental conditions did not have any paw swelling ( Figure 4A and B). Arthritic and radiological index There were significant reductions in arthritic indices (P # 0.05-0.001) at all dose levels for the extracts and reference drugs compared with the CFA-arthritic animals, except for the 100 and 300 mg/kg doses of ET PSE (Table 3). This trend of results was also seen in the radiological indices recorded. Changes in body weight recorded shows significant (P # 0.05-0.01) increments in only the 30 and 100 mg/kg AQ PSE-treated and diclofenac-treated arthritic animals. Hematological assessment Hematological assessment indicated significantly (P # 0.05-0.01) low white blood cell counts only in animals treated with 30 mg/kg AQ PSE or methotrexate, as well as normal animals kept under experimental conditions ( Table 4). The red blood cell count revealed a significant (P # 0.01) reduction in methotrexate-treated animals, but there was no significant difference (P . 0.05) in hemoglobin levels at all dose levels of the extract and reference drug-treated groups compared to the CFA-arthritic animals ( Table 4). Erythrocyte sedimentation rate and C-reactive protein levels were significantly (P # 0.01-0.001) low at all dose levels, except for the AQ PSE 300 mg/kg and ET PSE 100 and 300 mg/kg doses. Histopathological assessment of paw tissue Untreated CFA-induced arthritis showed an intense infiltrate of lymphocytes with foci of necrosis, pus collection, and scattered neutrophils in the ipsilateral paws, and chronic inflammatory changes dominated by lymphocytes with scattered plasma cells and foci of fibrosis with distortion of joint architecture in the contralateral paws ( Figure 5). Histological assessment of the ipsilateral paws of rats treated with AQ PSE 30 mg/kg showed no significant inflammatory changes. However, the contralateral paws showed only foci of mild chronic inflammatory change characterized by fibrosis with scattered lymphocytes. Tissue in the ipsilateral paws of the animals treated with AQ PSE 100 and 300 mg/kg showed chronic inflammatory changes dominated by lymphocytes with scattered plasma cells. However, the by fibrosis with scattered lymphocytes, but the contralateral paw showed no significant inflammatory changes. Diclofenac treatment showed foci of moderate chronic inflammatory change characterized by fibrosis with scattered lymphocytes in the ipsilateral paws, but there were no significant inflammatory changes in the contralateral paws. Dexamethasonetreated rats showed no significant inflammatory changes in the ipsilateral paws, but the contralateral paws showed moderate chronic inflammatory changes dominated by lymphocytes with scattered plasma cells and foci of fibrosis. Rats with incomplete arthritis induced by IFA showed mild chronic inflammatory changes dominated by lymphocytes with scattered plasma cells in the ipsilateral paw, while the contralateral paw showed no significant inflammatory changes. The experimental conditions did not result in any inflammation in the animals during the study period. Illustrations of these histopathological descriptions are shown in Figure 6. Discussion Adjuvant-induced arthritis in rats is recommended as a convenient model for preclinical studies of drugs used in the treatment of human arthritis, and has often been used Table 3 records of arthritic index, radiological index, and changes in body weight in CFA-induced and iFA-induced arthritic animals, CFA-induced arthritic animals treated with extracts and reference drugs, and normal animals kept under experimental conditions to study the mechanism of action and preventive effects of a number of disease-modifying antirheumatic drugs. 17,24,25 The development of adjuvant-induced arthritis in the rat can be divided into three phases, just like human rheumatoid arthritis, starting with the induction phase without evidence of synovitis, followed by early synovitis, and finally late synovitis with progressive joint destruction. [26][27][28] A good antirheumatic agent should be able to suppress one or more of these phases. This study has demonstrated that curative oral treatment of Sprague-Dawley rats using an aqueous or an ethanolic extract of P. stratiotes has potent antiarthritic properties in adjuvant-induced arthritis. All doses of the aqueous extract and the lower dose of the ethanolic extract suppressed joint inflammation significantly and ultimately reduced destruction of the joints, but only the lower dose of the aqueous extract prevented the systemic spread of arthritis. Joint protection and suppression of synovitis are known to be the ultimate goals of treatment of arthritis. 24,29 The extracts used in this study achieved these goals with effects similar to that of the reference drugs. The reference drugs, ie, methotrexate, dexamethasone, and diclofenac achieved inhibition of adjuvant-induced arthritis and prevented any spread of arthritis, which is consistent with the available literature. [30][31][32] Methotrexate, a disease-modifying antirheumatic drug and immunosuppressant, was used for comparison because it is a commonly prescribed front-line treatment for rheumatoid arthritis and the gold standard against which other systemic medications are compared. 31 Dexamethasone is known to inhibit the release of proinflammatory cytokines (tissue necrosis factor-α and interleukin-1β), which are known to play a central role in the propagation of the disease process in rheumatoid arthritis. 30. The anti-inflammatory effect of diclofenac is mediated mainly through inhibition of cyclo-oxygenase and prostaglandin production. 33. Rheumatoid arthritis is associated with weight loss and loss of lean body mass, known as rheumatoid cachexia. Rheumatoid cachexia is thought to be the end result of cytokine-driven hypermetabolism and is a key comorbidity in rheumatoid arthritis. 34,35 The loss of lean body mass is associated with decreased physical activity, muscle strength, and endurance in performing activities of daily living. 36 A loss greater than 40% of existing lean body mass often results in death. 37 Weight gain observed upon physical assessment of arthritic animals treated with AQ PSE 30 and 100 mg/kg and diclofenac, therefore, is ample evidence of good management of rheumatoid cachexia, minimizing the risk of mortality. The significantly low levels of serum C-reactive protein and erythrocyte sedimentation rate in the AQ PSE 30 and 100 mg/kg, ET PSE 30 mg/kg, and all reference drug-treated arthritic rats indicate remission of inflammation. Serum C-reactive protein is a sensitive but nonspecific marker of inflammation that responds rapidly to changes in underlying inflammatory disease activity, making its measurement an important tool for the detection and monitoring of inflammatory disease. 38 The erythrocyte sedimentation rate is a test that measures inflammation in the body indirectly. It measures the rate of settling or sedimentation of red blood cells in a capillary tube. Proteins produced during inflammation cause erythrocytes to move closer and stack up in a group. When this happens, they become denser and settle faster. The closer and faster the erythrocytes settle, the higher the value of the erythrocyte sedimentation rate. 39 Further, low levels of white blood cells indicate that the 30 mg/kg dose of AQ PSE (with the same effect seen for methotrexate treatment) is a potent antiarthritic treatment, given that elevated white blood cell levels are associated with active inflammation. 40 Radiographs are necessary to verify proper remission of disease and for accurate assessment of disease status. The measurement of paw or joint swelling gives an indication of edematous changes in this region, but the actual damage takes place in the tibiotarsal joint. 41 Reduction in bone configuration and increased bone resorption are the causes of bone loss in adjuvant-induced arthritic rats. [42][43][44] The X-rays clearly show that the aqueous extract and a low dose of the ethanolic extract of P. stratiotes decreased bone loss, even in cases of insignificant edematous changes in the contralateral paws, and therefore reduced bone degradation in arthritis. Histopathological studies of the paws strengthen the evidence of complete resolution of arthritis with AQ PSE 30 mg/kg despite evidence of pathology in the arthritic animals treated with the reference drugs. The antiarthritic effect of the aqueous and ethanol leaf extracts of P. stratiotes established in this study could be attributable to the presences of flavonoids, alkaloids, and sterols detected after phytochemical screening of the extracts. This assertion is supported by reports indicating that the presence of many biologically active phytochemicals, such as triterpenes, flavonoids, alkaloids, steroids, tannins, and glycosides, in various plant extracts may be responsible for their pharmacological properties. 27,[45][46][47][48] submit your manuscript \| www.dovepress.com Conclusion The study established that the aqueous and ethanolic extracts of P. stratiotes have antiarthritic activity in Sprague-Dawley rats with adjuvant-induced arthritis. The aqueous extract had better activity than the ethanolic extract, and warrants further study.	2017-06-27T03:39:57.591Z	2012-03-19T00:00:00.000	{ "year": 2012, "sha1": "8ce3edc5f01cfb63f5421e72ab472277675a116d", "oa_license": "CCBYNC", "oa_url": "https://www.dovepress.com/getfile.php?fileID=12341", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "4cbec226d5e87056adaf8153b15fcedf6d190004", "s2fieldsofstudy": [ "Biology", "Environmental Science", "Medicine" ], "extfieldsofstudy": [ "Medicine", "Chemistry" ] }

Subsets and Splits

No community queries yet

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